Xiaqing Pan successfully passed his Qualifying exam on Jan 19th, 2016. The title is “Feature extraction and content-based retrieval of 2D shapes and 3D mesh models”. His Qual Exam Committee includes: C.-C. (Jay) Kuo (chair), Antonio Ortega, Justin P. Haldar, B. Keith Jenkins, Aiichiro Nakano (outside member).
In this proposal, a robust two-stage shape retrieval (TSR) method is proposed to address the 2D shape and 3D shape (mesh model) retrieval problem. A challenge existing in most state-of-the-art retrieval methods is that they may retrieve globally dissimilar shapes in high ranks. He decomposed the decision process into two stages. He first removed globally irrelevant shapes in Stage I and then ranked the globally irrelevant shapes in Stage II. He designed more robust global features for each problem. The retrieval performances were significantly improved by integrating our global features with the proposed TSR method.