Chinese Character Segmentation

Tiang, King Ming (2015) Chinese Character Segmentation. Project Report. UTeM, Melaka, Malaysia. (Submitted)

Text (24 Pages) Chinese Character Segmentation 24 Pages.pdf - Submitted Version Download (180kB)

Abstract

This project is develop a segmentation algorithm to segment Chinese character. In order to develop an effective algorithm is necessary to study the existing segmentation algorithm/technique, including the history of segmentation technique and current segmentation technique. ut, there is the problem for segment Chinese character because of difficultly to identify the character’s radical of character itself, since some Chinese character can be the radical of other Chinese character. Besides that, handwriting also a problem, because of handwritten is difficulty be recognize. Handwritten character due to different writes has their own style of writing therefore the feature of handwritten character is not constantly same, although it is written by same writer, let alone the difference writer. This project develop a segmentation algorithm using the concept of image processing like binarization and therolding to convert image into binary image. After that convert the image coordinate into horizontal and vertical histogram to identity the point for segmentation of image. After the image be segmented it will be featurize by using triangle geometric feature. Then, system will calculate the distance between the test’ feature data and model data, and the distance between each character will using MAP(Mean Average Precisions) to calculate result and their ranking. In conclusion, this application consist potential for future technology. But the application still has a lot of improvement to be done in future.

Actions (login required)

View Item