The invention discloses a variation model-based
single image defogging method, which comprises the following steps of 1, acquiring an image in the foggy weather, namely a to-be-restored image I; 2, acquiring an initial
transmittance image t0(x) based on the dark-channel prior theory; 3, according to an
image restoration model in the foggy weather, acquiring an initial restored image J0(x) in the condition of a known
transmittance; 4, according to the acquired initial
transmittance image t0(x), calculating an initial inverse-edge image M0; 5, based on the inverse problem solving idea and a variation model, constructing an energy function F(t, J) for the
single image restoration; 6, based on the step-by-step
gradient descent flow method, minimizing the energy function F(t, J); 7, at the ending of the
iteration process, extracting an image Jn(x) as a restored image, and an image tn(x) as a transmittance image. Based on the above method, the homogeneity of the transmittance can be maintained. Meanwhile, the defect that the
image edge information goes wrong can be overcome.