The state-of-arts global thresholding techniques are fast and efficient to convert the gray scale document image into a binary image. However, they are unsuitable for complex and degraded documents. Moreover, global thresholding techniques produce border noise when the illumination of the document is not uniform. Other methods that depend on local thresholding techniques are efficient in the case of degraded document images, but have common disadvantages include the dependence on the region size and the image characteristics, and the computational time. In this paper we propose a method to overcome the limitations of the related global and local threshold techniques by using the concept of Retinex theory based on Median filter which can effectively enhance the degraded and poor quality document image. High quality results in terms of visual criteria and OCR performance is produced compared to the previous works. © 2014 IEEE.