A Sobel gradient histogram equalization(GHE) algorithm is proposed to enhance the contrast of infrared images during dynamic range compression. In contrast to previous histogram equalization(HE) methods, this method adaptively assigns a high contrast to the strongly graded parts of the image, preserving and enhancing more details in the 16-bit image. Dual Gamma mapping is then used to adjust the mapping curve to effectively suppress overexposure in the bright parts of the image while improving the detail in the shadows. Compared with the traditional histogram equalization algorithm, this method has better effects on dark area detail processing, overexposure suppression, and contrast enhancement.