摘要
甲骨拓片上甲骨文字定位及分割是甲骨文识别的关键,定位的好坏将直接影响到拓片上甲骨文字的识别率。为了尽可能准确地将甲骨拓片从复杂的噪声中分离出来,首先对原始拓片进行选取合适的阈值实现对文字的粗略分割,然后运用形态学方法进行细定位。通过对《甲骨文合集》中的甲骨拓片图像实验数据进行基于阈值分割及形态学处理验证表明,该方法能够较好地实现拓片文字的定位。
The location and segmentation of rubbing text of oracle bone inscriptions are the keys for oracle recognition. The positioning will directly affect the recognition rate of oracle bone inscriptions. In order to accurately separate characters from the background noise, first, along with the preprocessing of the original oracle bone inscriptions, threshold segmentation is done to achieve a rough segmentation of the text regions. Then based on morphology, the fine location of the text regions is realized. The data of the rubbing collection of the oracle bone inscriptions are processed based on edge and mathematical morphology. The results of experiments show that this method can better realize the location of rubbing text.
出处
《北京信息科技大学学报(自然科学版)》
2014年第6期7-10,24,共5页
Journal of Beijing Information Science and Technology University
基金
国家自然科学基金项目(60875081)
关键词
甲骨拓片
阈值分割
数学形态学
文字定位
rubbing text of oracle bone inscriptions
threshold segmentation
mathematicalmorphology
text location