multi-modal remote sensing data for active push

Satellite network simulation setup: Use the Jilin-1 satellite as a reference to construct a network of 300 smart remote - sensing satellites. They're deployed in 20 sun - synchronous orbits (656km high, 98.04° inclination). The comms satellite constellation, inspired by Starlink's design, has 650 satellites in 25 Walker orbits (550km high, 53° inclination).Generate remote - sensing content based on the LS satellites' lat - lon positions at different time intervals. To be specific, create a square - centered area using the satellites' lat - lon coordinates. A total of 280 remote - sensing contents are generated during the simulation period.MTARSI, a high - resolution remote - sensing dataset from Hefei University, supplied the optical images. It has 9,385 images of 20 airplane types, like A - 10 and B - 1, from 36 airports in the US, UK, etc., under varying conditions, annotated by seven experts.For SAR images, the public SAR - ACD dataset was used. It covers 6 civil and 14 military/special - purpose airplane categories, such as A220 and ARJ21, with a 1 - meter resolution. Based on HH - polarized time - series SAR images, it focuses on areas like Shanghai Hongqiao Airport and is labeled via expert interpretation and optical image comparison. After data augmentation, the public part has 3,032 images of the 6 civil airplane categories, around 500 each, across multiple stages.By multi - scale reconstruction, a cross - modal image set with five resolutions (0.1m, 0.3m, 0.5m, 1m, 2m) was generated from both datasets. After sample balancing, the dataset includes 280 images.