For the binarization of text image captured under uneven illumination,the article propose a new methods which
combines gray value statistics with global threshold to remove uneven illumination effects.The method uses the gray
value statistics method to extract the background image, so that the original subtract the background. Finally to get
binary image by using improved simple statistical global threshold method. The experimental results show that this
method not only retains the advantages of global threshold such as simpleness and efficient, but also has the advantages
of the local threshold which retains the details of the character strokes well and avoids the disconnection and
shade.Thereby the method improves the recognition efficiency of the binary text.