Fast Adaptive Compensation based Binarization Method for Text Images
Qi Liu · Jisuanji fangzhen · 2011
Image binarization plays a key role in the text image processing system.Traditional binarization algorithms cannot provide reliable results.In this paper,we proposed a method which estimated the intensity of the background of the text image and adaptively adjusted the gray-scale of the original image based on the estimated background.On this adaptively adjusted image,user can achieve a satisfied binarization result using a simple global thresholding based method such as Otsu.The experimental result shows that our method outperforms the other algorithms for the uneven illuminated text images.The accuracy of the digital recognition for the car plate image is also improved using the proposed binarization method.