Offline Writer Verification Based on Bangla Handwritten Characters Using Enhanced Textual Feature
Jaya Paul, Anasua Sarkar, Chayan Halder, Kaushik Roy · 2017
This chapter proposes a novel offline text dependent writer verification method based on Gray Level Co-occurrence Matrix (GLCM). The method is used to combine the result of Dual Tree Complex Wavelet Transforms (DT-CWT) and GLCM features to enhance the GLCM feature performance. The texture is a very important characteristic, which is used to verify the known writer from unknown writers. We have used 615 documents from 123 writers producing 43665 numbers of Bangla Isolated characters. The results show that these selected texture features have high distinction accuracy and hence is appropriate for efficient use in the challenging task of automatic writer verification.