Title :
Handwritten character recognition using multi-resolution histograms of oriented gradients
Author :
Qinyunlong ; Sunguangling
Author_Institution :
Sch. of Commun. & Inf. Eng., Shanghai Univ., Shanghai
Abstract :
In this paper, a novel handwritten character recognition scheme by using multi-resolution histograms of oriented gradients (MHOG) is proposed. Histograms of oriented gradients of multi-resolutions are fused with canonical correlation analysis (CCA). The scheme is evaluated on three handwritten character databases and the result has demonstrated the effectiveness of the presented scheme.
Keywords :
correlation methods; gradient methods; image recognition; image resolution; canonical correlation analysis; handwritten character recognition; multiresolution histograms; oriented gradients; Character recognition; Data mining; Databases; Feature extraction; Fuses; Fusion power generation; Histograms; Image resolution; Multiresolution analysis; Sun; canonical correlation analysis; handwritten character recognition; multi-resolution histograms of oriented gradients;
Conference_Titel :
Communication Technology, 2008. ICCT 2008. 11th IEEE International Conference on
Conference_Location :
Hangzhou
Print_ISBN :
978-1-4244-2250-0
Electronic_ISBN :
978-1-4244-2251-7
DOI :
10.1109/ICCT.2008.4716216
