Deity face recognition using schur decomposition and hausdorff distance measure
Ramadoss Balakrishnan, Rajkumar Kannan, Suresh Kannaiyan, Sridhar Swaminathan · 2013
Face recognition in deity images is a challenging problem. Most of the existing face recognition methods are very sensitive to pose and illumination changes. This paper proposes a new technique for deity face recognition which is suitable for pose and illumination changes. The proposed approach uses Schur decomposition to speedup PCA computations and doubly modified Hausdorff distance for measuring similarity between different face edge maps. In addition, this paper introduces a new dataset named as Indian DEity dataSet (IDES) for face recognition which contains a collection of face images of Indian deities. Performances of the proposed method for deity face recognition are experimented with IDES dataset. The results show that the proposed method provides promising face recognition accuracy with pose and illumination changes.