Character segmentation and restoration of Qin-Han bamboo slips using local auto-focus thresholding method
Songxiao Cao,Zichao Shu,Zhipeng Xu,Dailiang Xie,Ya Xu
DOI: https://doi.org/10.1007/s11042-022-11988-z
IF: 2.577
2022-02-01
Multimedia Tools and Applications
Abstract:This paper presents a novel auto-thresholding method for character segmentation and restoration of historical Chinese documents. The objective was to segment and restore the characters of Qin-Han bamboo slips effectively with complex background noise. To that end, giving a whole page image with several bamboo slips, the proposed method first extracted and straightened every single slip by connected component analysis. Furthermore, for every straightened slip, a horizontal histogram projection method was used to segment all character regions. After that, a novel auto thresholding method, which was motivated by the auto-focus process of camera, was used to find the optimal threshold of every character region. In this method, the algorithm traversed all the thresholds in a certain range and generated an Effective Character Contour Length (ECCL) value for each threshold, then multi-Gaussian model was used to fit the ECCL curve and the global peak position of ECCL curve was the needed final optimal threshold for the character region. Experimental results showed that the proposed method was effective for historical character segmentation and restoration under complex background noise. Compared to five existing state of the art algorithms, including Otsu, integral image adaptive thresholding method, Sauvola, GAN denoising and SAE algorithm, the proposed method can not only restore the whole characters more completely, but also suppress the noise better.
computer science, information systems, theory & methods,engineering, electrical & electronic, software engineering