International Journal on Advanced Science, Engineering and Information Technology, Vol. 9 (2019) No. 6, pages: 2216-2222, DOI:10.18517/ijaseit.9.6.9001

Research and Development of Feature Extraction from Myanmar Palm Leaf Manuscripts for the Myanmar Character Recognition System

Nwe Nwe Soe, Win Htay


This paper proposed Myanmar palm leaf manuscript handwriting OCR system. Each text area in the Myanmar palm-leaf manuscript is segmented. This segmented character text image is needed to be recognized to transform to Myanmar handwritten characters which express Myanmar’s precious historical and invaluable information. This paper involves two essential steps: preprocessing and feature extraction. The preprocessing is carried out to extract the attractive palm-leaf manuscript region from the Images automatically are taken by the camera and to support the enhanced images for subsequence processes of Myanmar character recognition from Myanmar palm leaves. The one-dimensional segmentation approach is used to crop leaf area in the image which is taken with high resolution. Line count analysis is also done to extract the region for using enough line count. After that, line segmentation is carried out using Object Frequency Histogram along the horizontal lines which can find the best optimal points between the lines. Similarly, the same technique but vertically is used to get each character or smallest group of characters. Totally 18 features are extracted to recognize the Myanmar palm-leaf manuscript characters. Although the experimental results are good enough but some difficulties are still needed to take account related to the connected components. 


Myanmar palm leaf manuscripts; feature extraction; OCR; line segmentation; character segmentation.

Viewed: 253 times (since Sept 4, 2017)

cite this paper     download