The image contrast and definition of remotely sensed image are the important factors of image quality. This paper presented a novel self-adaptive enhancement method for contrast and definition. Firstly,nonlinear transformation is operated on the histogram of an input image to reduce the distribution range of the image,and then linear stretch is used to expand the grey level to full permitted range. Experiments and comparison indicate that the presented method can improve the contrast and the definition with a little loss of information entropy and the enhanced results is rather fine than current main method of enhancement. Also,the presented method can be used to both panchromatic and colorful remotely sensed image.