Remote sensing image defogging method based on generative adversarial network

• Main Authors: JIA YANFEI, SHEN HELIN
• Format: Patent
• Language: Chinese; English
• Published: 31.05.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

A remote sensing image defogging method based on a generative adversarial network belongs to the technical field of computer vision, and comprises the steps of constructing a color and illumination feature extraction network branch, constructing an edge feature extraction network branch, constructing a feature extraction backbone network branch, extracting image shallow layer features by adopting conventional convolution, and performing feature fusion and image recovery. And constructing an adversarial network, measuring the performance of the remote sensing image defogging network by using a generative adversarial network loss function, and minimizing the loss function by using a gradient optimization method so as to realize optimization of network parameters and obtain final network parameters. According to the method, the problems of illumination distortion, color distortion and local blurring of an image obtained by an existing remote sensing image defogging technology are solved; while the definition of