Apriori Data Mining on Rotationally Invariant Multiresolutional Moments for Pattern Recognition
Annupan Rodtook, Stanislav S. Makhanov · Advances in intelligent systems research/Advances in Intelligent Systems Research · 2006
We propose a new feature selection procedure based on a combination of a pruning algorithm, Apriori mining techniques and fuzzy C-mean clustering.The feature selection algorithm is designed to mine on a multiresolution filter bank composed of rotationally invariant moments.The numerical experiments, with more than 10,000 images, demonstrate an accuracy increase of about 5% for a low noise, 15% for an average noise and 20% for a high-level noise.