k-NN classification of handwritten characters via accelerated GAT correlation
作者:
Highlights:
• ●We enhance k-NN classification via a distortion-tolerant GAT correlation technique.●We accelerate GAT correlation using separation of variables and lookup tables.●We examine both shape matching and discrimination abilities of GAT correlation.●GAT correlation is far superior in recognition accuracy to the tangent distance.●Computational cost of GAT correlation is reduced to about 6% of the original cost.
摘要
●We enhance k-NN classification via a distortion-tolerant GAT correlation technique.●We accelerate GAT correlation using separation of variables and lookup tables.●We examine both shape matching and discrimination abilities of GAT correlation.●GAT correlation is far superior in recognition accuracy to the tangent distance.●Computational cost of GAT correlation is reduced to about 6% of the original cost.
论文关键词:Affine-invariant template matching,k-NN classification,Normalized cross-correlation,Handwritten character recognition
论文评审过程:Available online 16 May 2013.
论文官网地址:https://doi.org/10.1016/j.patcog.2013.05.005