Detection of Single and Multi-character Tulu Text Blocks

Segmentation of line into words, characters is another important exercise of OCR pipeline. In south Indian handwritten script named Tulu, it is an imperative research problem as it involves complex structural characteristics. The degree of processing aggravates even more in the presence of touching/overlapping character blocks along with usual operational intricacies. In the perspective of optimizing the touching/overlapping character segmentation of Tulu script, an intermediate processing step to simplify the segmentation process has been introduced in this work based on text block characteristics.