液晶与显示, 2017, 32 (4): 287, 网络出版: 2017-05-03
基于计算机视觉的X射线图像异物分类研究
X-ray image illegal object classification based on computer vision
X射线异物 Contourlet 变换 Tamura纹理特征 随机森林 分类 illegal object with X ray Contourlet transform Taruma texture feature random forest classification
摘要
在安检领域, 目前最主要的手段是人工分析X光图像, 以检测是否隐藏的违禁品。由于人工检测存在较强的主观性, 并且在安检员疲劳时容易造成漏判、错判。针对这一问题, 对X光异物图像进行自动识别研究, 提出了基于Tamura纹理特征和随机森林的X射线异物分类方法。介绍了基于计算机视觉的X光违禁品自动检测识别系统; 提出一种基于Contourlet变换的Taruma纹理特征提取方法, 通过该方法得到Taruma纹理特征向量; 最后采用随机森林分类器对违禁品图像进行分类判断。实验结果表明, 基于Tamura纹理特征和随机森林的X射线异物分类能够有效地区分不同种类的违禁品。
Abstract
In the field of security, manual X-ray images analysis is a common method to detect illegal object. Due to manual analysis’ strong subjectivity, it is easy to cause false detection. The illegal object classification by Tamura texture feature and random forest with X ray is proposed in this paper. Firstly, an automatic illegal object detection and identification system based on computer vision is introduced. Then a method based on Contourlet transform is proposed to extract Taruma texture. Finally, the random forest classifier is adopted to classify illegal objects. Experimental results show that the illegal object classification by Tamura texture feature and random forest with X ray can effectively distinguish different kinds of illegal object.
王宇, 邹文辉, 杨晓敏, 姜维, 吴炜. 基于计算机视觉的X射线图像异物分类研究[J]. 液晶与显示, 2017, 32(4): 287. WANG Yu, ZHOU Wen-hui, YANG Xiao-min, JIANG Wei, WU Wei. X-ray image illegal object classification based on computer vision[J]. Chinese Journal of Liquid Crystals and Displays, 2017, 32(4): 287.