Generative confrontation cloud simulation generation method considering feature redundancy

• Main Authors: LI WENZHUO, SUN KAIMIN, LI WANGBIN
• Format: Patent
• Language: Chinese; English
• Published: 12.08.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention discloses a generative adversarial cloud simulation generation method considering feature redundancy, which is characterized in that a generative adversarial network used for simulating cloud is set, the generative adversarial network comprises a discrimination network and a generative network, the generative network comprises an encoder and a decoder, and a feature channel selector is embedded in the decoder; stacking the base map and the cloud mask, inputting the stacked base map and cloud mask into an encoder part in the generative network, and processing the input image and cloud mask to generate a high-dimensional feature; generating a decoder part in the network, and embedding a feature channel selector module on each feature size to screen features; inputting a coarse result generated by the generative network into a discrimination network to obtain an authenticity determination result of the simulation result; and after the training is completed, carrying out cloud simulation by using an