Remote sensing image data enhancement is a very important preprocessing step in imaging system for agriculture applications. In this paper, after demonstrating the equivalence of the sharpening by the Laplacian to inverse heat equation processing, a robust inverse diffusion equation is presented, which sharpens image details by a robust Laplacian. Image gradient magnitude is used to avoid the noise magnification. At the same time, the minmod function is used to control diffusion flux adaptively, which eliminates effectively overshoots inherent in the Laplacian. Experimental results demonstrate that this algorithm can enhance important details of image data effectively without overshoots, affording the opportunity for a good interpretation and subsequent processing.