Object classification in 3D baggage security computed tomography imagery using visual codebooks
作者：
Highlights：
• Novel investigation of BoW model for object classification in 3D baggage CT scans.
• Four descriptor types and three codebook assignment methodologies compared.
• Simple density-based descriptors outperform more complex descriptors.
• Optimal true and false positive rates for classification of handguns and bottles.
• Low resolution, noise and artefacts limit performance.
摘要
Highlights•Novel investigation of BoW model for object classification in 3D baggage CT scans.•Four descriptor types and three codebook assignment methodologies compared.•Simple density-based descriptors outperform more complex descriptors.•Optimal true and false positive rates for classification of handguns and bottles.•Low resolution, noise and artefacts limit performance.
论文关键词：3D Object classification,Bag of (Visual) Words,3D descriptors,SIFT,RIFT,Baggage-CT
论文评审过程：Received 19 February 2014, Revised 9 January 2015, Accepted 7 February 2015, Available online 14 February 2015.
论文官网地址：https://doi.org/10.1016/j.patcog.2015.02.006
