Rough-Bayesian approach to select class-pair specific descriptors for HEp-2 cell staining pattern recognition
作者：
Highlights：
• The proposed method selects a set of relevant descriptors for each pair of classes.
• Final feature set for multiple classes is formed considering all pairs of classes.
• decision theory and rough set are used to assess the relevance of descriptors.
• The efficacy of the proposed method is demonstrated on several HEp-2 cell databases.
• Significant increase in accuracy is noted employing class-pair specific descriptors.
摘要
•The proposed method selects a set of relevant descriptors for each pair of classes.•Final feature set for multiple classes is formed considering all pairs of classes.•decision theory and rough set are used to assess the relevance of descriptors.•The efficacy of the proposed method is demonstrated on several HEp-2 cell databases.•Significant increase in accuracy is noted employing class-pair specific descriptors.
论文关键词：HEp-2 cell images,Staining pattern recognition,Texture analysis,Rough sets,Bayes decision theory
论文评审过程：Received 30 September 2019, Revised 13 March 2021, Accepted 31 March 2021, Available online 8 April 2021, Version of Record 22 April 2021.
论文官网地址：https://doi.org/10.1016/j.patcog.2021.107982
