Offline Persian Writer Identification Based on Wavelet Analysis
Mohammad Kazem Akbari, Reyhaneh Eslami, Mostafa Haghi Kashani, Shahr-e-Qods Branch, Mostafa Haghi Kashani · 2012
In this paper we introduce a new efficient approach for writer identification and verification based on written style. At first, the handwritten text image is normalized then the handwritten features are extracted by wavelet transform and KNN1 co-occurrence matrix. At last, handwritings are classified by a classifiers and the writer is being identified. Experimental results on variety of handwriting databases confirm the efficiency of this method. Writer recognition rate of this method is 93.3%.Numerical results are presented in continue.