Binarization is one of the chief technology in digital image processing in library, The quality of the digital books lie on the result of the binarization. This paper introduces a new whole binarization method by using image gray statistical characteristic value as threshold. In the course of the text image processing, the algorithm can automatically realize the threshold segmentation by using the characters of the high-low pass filter based on the histograms of distinct text images. The algorithm has lower time complexity and space complexity, and has been applied in one digital library system.