The invention discloses a foggy weather
image enhancement method based on
fractional differential and dark channel prior. The method herein includes the following steps: 1. inputting a foggy weather image I, conducting dark channel prior and
Retinex algorithm processing on the I, obtaining an initial de-foggy image J (x,y); 2. segmenting the J (x,y) to a foreground image J1(x,y) and a background J2(x,y); 3. separately computing the optimal
fractional differential order number v1 corresponding to J1(x,y) and the optimal
fractional differential order value v2 corresponding to J2(x,y); 4. determining a
mask coefficient and a
mask size, constructing a fractional
differential operator mask w (s,t); 5. separately introducing the fractional differential
order number v1 and fractional differential order value v2 which are obtained from 3 to the w (s,t), obtaining w1(s,t) and w2(s,t), conducting
convolution operation on the pixel points of w1(s,t) and J1(x,y), and conducting
convolution operation on the pixel points of w2(s,t) and J2(x,y); and 6. outputting the image of I which after the
image enhancement. According to the invention, the method herein addresses poor de-foggy effects which are often caused in the process of enhancing foggy images by using fractional
differential algorithm which has single fractional differential order in the prior art.