It is established that perceived image contrast depends on viewing conditions, especially the level of surround illumination. In this research, perceived image contrast was re-measured and modeled. The input image was rendered from XYZ values through an image appearance model (iCAM06), in which the default viewing condition is dark surround. The image physical contrast was controlled by an exponential factor in the lightness channel (I) in the IPT space. The paired comparison psychophysical method was used to measure the perceived equivalent image contrast under different levels of surround illumination. The corresponding surround compensation gamma ratio was measured by normalizing all physical gamma values to the one with surround luminance with 100% of image white. The changes in perceived image contrast with different levels of surround illumination were modeled as a power function with different exponent between the perceived lightness and relative physical luminance level in the image. The exponential ratio in this research for average, dim, dark surround is different from the classical result from Bartleson & Breneman or Hunt, CIECAM97s, CIECAM02, and some other research results. An optimal ratio of 1.0:1.10:1.14 was found as opposed to the traditional values of 1.0:1.25:1.50. The results also show that image content plays a very important role when people view an image and judge the perceived image contrast. The results from this research can be used to improve the surround compensation factor in current image and color appearance models.