Hausdorff distance with k-nearest neighbors

Authors:
Jun Wang;Ying Tan
Affiliations:
Key Laboratory of Machine Perception (Ministry of Education), Peking University, P.R. China,Department of Machine Intelligence, School of EECS, Peking University, P.R. China;Key Laboratory of Machine Perception (Ministry of Education), Peking University, P.R. China,Department of Machine Intelligence, School of EECS, Peking University, P.R. China
Venue:
ICSI'12 Proceedings of the Third international conference on Advances in Swarm Intelligence - Volume Part II
Year:
2012

Citing 9
Cited 0

Efficient Visual Recognition Using the Hausdorff Distance

Efficient Visual Recognition Using the Hausdorff Distance
Face Recognition Using Line Edge Map

IEEE Transactions on Pattern Analysis and Machine Intelligence
Comparing Images Using the Hausdorff Distance

IEEE Transactions on Pattern Analysis and Machine Intelligence
Pattern Classification (2nd Edition)

Pattern Classification (2nd Edition)
Hybrid image matching combining Hausdorff distance with normalized gradient matching

Pattern Recognition
Elastic shape-texture matching for human face recognition

Pattern Recognition
A new Hausdorff distance for image matching

Pattern Recognition Letters
An incremental Hausdorff distance calculation algorithm

Proceedings of the VLDB Endowment
Object matching algorithms using robust Hausdorff distance measures

IEEE Transactions on Image Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Hausdorff distance (HD) is an useful measurement to determine the extent to which one shape is similar to another, which is one of the most important problems in pattern recognition, computer vision and image analysis. Howeverm, HD is sensitive to outliers. Many researchers proposed modifications of HD. HD and its modifications are all based on computing the distance from each point in the model image to its nearest point in the test image, collectively called nearest neighbor based Hausdorff distances (NNHDs). In this paper, we propose modifications of Hausdorff distance measurements by using k-nearest neighbors (kNN). We use the average distance from each point in the model image to its kNN in the test image to replace the NN procedures of NNHDs and obtain the Hausdorff distance based on kNN, named kNNHDs. When k=1, kNNHDs are equal to NNHDs. kNNHDs inherit the properties of outliers tolerance from the prototypes in NNHDs and are more tolerant to noise.