A Selective Local Thresholding Method for Camera Document Image

This paper presents a selective local thresholding method of a camera based document image for applying character segmentation. It is very difficult to binarize a camera based document image due to inconsistent lighting condition. It makes difficulties of character extraction and recognition. A selective local thresholding method in character region can reduce illumination effect and extract properly characters. Edge operator is used to detect some candidate character. The local thresholding method is applied in the selected candidate character region. Distribution of intensity value of pixels is used to binarize the character region. It is an efficient method to binarize camera image and to extract characters from background image. The proposed method can be implemented in mobile device because of high speed processing time and low memory requirement. To evaluate the proposed method, we have experimented with camera document image of the ETRI database. An encouraging results have been obtained. Key-Words: Camera document recognition; Local thresholding; Character segmentation