Electronics and Telecommunications Research Institute

Soo Hyun Cho

Won Pil Yu

Kye Kyung Kim

Su Young Chi

Yun Koo Chung

Hyoung Gu Lee

Young Sup Hwang

Disclosed is a method of segmenting handwritten touching numeral strings having a non-vertical segmentation line. The method comprises the steps of: receiving an image of the handwritten touching numeral strings having the non-vertical segmentation line extracted from a specially patterned document; smoothing an uneven shape of the image of handwritten touching numeral strings; searching a contour line for the image of smoothed handwritten touching numeral strings; searching a candidate segmentation region, in which the touching numeral strings may exist, from the searched contour line; searching a candidate segmentation point within the candidate segmentation region; searching a vertical segmentation line for segmenting the handwritten touching numeral strings from the candidate segmentation point; searching the non-vertical segmentation line for segmenting the handwritten touching numeral strings having any inclination from the candidate segmentation lines; and segmenting the handwritten touching numeral strings by use of the vertical and non-vertical segmentation lines.

Method of segmenting handwritten touching numeral strings having non-vertical segmentation line

Disclosed is a method of segmenting touching numeral strings contained in handwritten touching numeral strings, and recognizing the numeral strings by use of feature information and recognized results provided by inherent structure of digits. The method comprises the steps of: receiving a handwritten numeral string extracted from a pattern document; smoothing a curved numeral image of the handwritten numeral string, and searching connecting components in the numeral image; determining whether or not the numeral string is a touching numeral string; if it is determined that the numeral string is the touching numeral string, searching a contour of the touching numeral string image; searching candidate segmentation points in the contour, and segmenting sub-images; computing a segmentation confidence value on each segmented sub-image by use of a segmentation error function to select the sub-image with the highest segmentation confidence value as a segmented numeral image in the touching numeral string image; if it is determined in the step c that the numeral string is not the touching numeral string, extracting a feature to recognize the segmented numeral image; segmenting the numeral image selected from the touching numeral string in the highest segmenting confidence value; and obtaining remaining numeral string image.

Method for segmentation-based recognizing handwritten touching numeral strings

Growing community of inventors

Taejon, South Korea

Soo Hyun Cho