Abstract: In order to tackle color distortion and appearance blur of leaf images caused by atmospheric dust, in this paper, an automatic algorithm is proposed of dust removal for a single image, based on an optical model. We construct an image degradation model according to the transmission way of light in the dust layer and atmosphere. The two model parameters, the atmospheric light and transmission map, are then estimated by the optical reflectance imaging model and dark channel prior. Based on the model, we can remove the effect of dust on the image quickly. The images of dusty Color Checker Chart and grape leaves are employed to check the effectiveness of the algorithm. The experimental results show that the high-quality dust-free images can be recovered, with the deviations between the H and S components of them and those of the clean images being significantly reduced. And the algorithm is applicable to the images under various conditions, i.e., weather, illumination and grape varieties.