Seal Extraction Based on Clustering and Local Thresholding Techniques
Cheng Yun Wang, You Chen · Advanced materials research · 2014
In this paper, we propose a method to extract seal imprints from bank document images based on color information. Firstly we use K-means clustering algorithm and local thresholding techniques to distinguish seal imprints from background areas. Secondly we wipe off printed characters and other noise upon the nearest neighbor classifier. Finally through a series of post-processing steps some missing pixels are added to the seal imprints. The result of our experiments proves that this method is capable of extracting Chinese seal imprints in most cases.