[ Johannes Niedermayer | Research | Publications | Services ]

Research

I am very interested in analyzing and querying visual 2D (image) and 3D (3D scan) data such as for object recognition and image retrieval, including the following fields:

• Feature extraction, indexing and matching - I am currently addressing these topics in an industry project. This includes:
  • Keypoint-based features, both real-valued (e.g. SIFT) and binary (e.g. ORB and BinBoost)
  • Quantization of these features for efficient indexing
  • Geometric verification
  • Deep learning for feature extraction and feature reduction (convolutional neural networks and auto-encoders)

• Analysis of 2D and 3D imagery

Beyond querying and analyzing visual data, my research interests include processing similarity queries in general databases; I was funded for this research until 2013:

• Nearest neighbor queries

• Reverse nearest neighbor queries

• Similarity queries for uncertain data