2R-8
Preliminary Study toward Anomaly Detection in Capsule Endoscopy Images based on Higher-order Local Auto Correlation
○胡 尓重（筑波大），野里博和，坂無英徳，村川正宏（産総研）
The aim of this research project is to develop the system automatically detecting anomaly (lesion, residue, blood etc.) in capsule endoscopy based on Higher-order Local Auto Correlation (HLAC). Toward this goal, this paper focus on the HLAC feature extraction, especially image preprocessing. The computational experiment show the results of some image preprocessing methods for HLAC feature extraction, and their effects on the performance of anomaly detection are examined.