Author(s): Shakunthal ABS, & CS Pillai
Handwritten text line segmentation is regarded as an important test in record picture evaluation. The difficulty in today's printed text, Skewed lines, curved lines contacting and covering components, generally words or characters, among lines, and geometrical features of lines. The difficulty involved in segmenting Handwritten Documents for Indian dialects are Telugu, Tamil, and Malayalam. Manually written archives with bended and non-parallel text lines also perform segmentation and recognition testing. This work considers text line segmentation of manually authored Kannada content records using ICA. In this preliminary step, the authors provided an improved approach for manually written content line division to the proposed technique. The proposed system includes improved text-line segmentation as well as skew estimation and the dataset is a handwritten Kannada document. The preprocessing approaches are: (i) Filtering, (ii) Grey scale conversion, and (ii) Binarization. The ESLD method is used to estimate distance between text lines and R Clustering aids in word grouping or the Connected Components. Skew estimate can also be achieved by determining the skew angle with respect to the gap. The output demonstrates that the proposed system out performs the competition.