Abstract: For unsighted detail and texture of images due to many factors during imaging, a novel technique of image enhancement by two-dimensional histogram with self-adaptive window was proposed based on two-dimensional histogram theory of Turgay Celik. This method considered image details and texture. By analyzing the relationship between two-dimensional histogram and texture characteristics and defect of fixed window, self-adaptive window was proposed by average grads. It was suited the region characteristics to construct two-dimensional histogram and accomplish enhancement through comparing the distributing with ideal one. Many experimental results and comparison show that this method can overcome badly adaptability, excessive enhancement, local weakening and other problems caused by fixed window. The enhanced images display clear texture and more pleasing visual content so that the method is useful for engineering.