"Qimingxing" has been in the sky for half a year and is about to give birth to "a picture"; LuoJia's "Gemini" is online, hoping to weave a "web".
On September 5th, at the Luojia Laboratory in Hubei Province, when it comes to technological innovation, professors Jin Guang and Hu Xiangyun told Hubei Daily reporters that "the more difficult it is, the more difficult it is to tackle." They are respectively involved in the research work of the "Qimingming-1" micro nano satellite and the specialized deep learning framework/database for remote sensing image intelligent interpretation - LuoJiaNET/LuoJiaSET, which is currently in progress.
The first self-developed satellite by students will be launched into space, and "One Map of the Province" is about to be released
On February 27th this year, the "Qimingming-1" micro nano satellite was launched into space. This "Qimingxing" is China's first visible light hyperspectral and night light multispectral multi-mode in orbit programmable micro/nano satellite. Its main research and development work is mainly carried out by students on campus, jointly completed by multiple units such as Luojia Laboratory and Wuhan University. It has the characteristics of low cost, fast response, and high flexibility.
"The satellite has been in the sky for half a year and is operating normally." In front of the satellite simulation system, a "small box" is soaring along the established orbit above the Earth. In the view of Jinguang, "Qimingxing" is a "space laboratory" where students can break through the limitations of textbooks and classrooms, design application plans according to their creativity, and fully familiarize themselves with satellite related technologies and applications through the development, launch, and monitoring of satellites. "At present, all maintenance work for satellites is completed by students, and teachers can almost ignore it," Jin Guang said with a smile.
"I am processing the image of Wu Rucuo in Xizang, and will double the image resolution on the existing basis to make it look clearer." Recalling the process of participating in the research and development of "Venus", Zhang Xuefeng, a doctoral student, said frankly that he was originally a software student, but he had to deal with many hardware when participating in the research and development of digital satellite systems, which made him difficult, "Every component needs to be understood and tested. Fortunately, with the guidance of teachers, a digital system was built in a short time. Time is tight and tasks are heavy, which has allowed me to grow quickly."
The acquisition of high-precision remote sensing images requires precise calculation of satellite positions. Professor Zhao Qile's team from the laboratory has conducted in-depth research and continuously broken through core technologies, improving the accuracy of China's satellite orbit determination from tens of meters to 1-2cm. The latest achievements have been applied to the SAR satellites Gaofen-3 02 and 03 launched in November 2021 and April this year, with a relative positioning accuracy of 1mm between the two satellites. This has laid an important foundation for improving China's remote sensing observation capabilities and achieving all-weather and all-weather observation of the Earth, To calibrate the heading and provide escort for obtaining high-precision satellite images.
It is understood that the laboratory's recent important achievement, "A Map of Hubei", is about to be released. According to Jinguang, the entire weight of "Qimingxing" is about 19kg. Although it is compact, it is equipped with visible light and infrared cameras. During the day, it can obtain 32 spectral bands of hyperspectral Earth surface images, and at night, it can obtain 8 spectral bands of night light Earth surface images and infrared daytime remote sensing images. "A map of Hubei" is a concatenation of hyperspectral images, which is equivalent to the "fingerprint" of the land of Hubei. It can provide important basis for water environment monitoring, crop growth trend monitoring, urban economic development, light pollution monitoring, and natural resource investigation in multiple fields.
"Whether the land under one's feet is forest or farmland, and whether corn or rice is planted in the farmland, can all be displayed through a 'map'. With this map, it is equivalent to having a 'fingerprint' of the land of Hubei, which will play an important role in environmental monitoring, protection, and governance of Hubei, and serving the local economy." Jin Guang said.
LuoJia's "Gemini" Achieves Autonomous and Controllable Remote Sensing Interpretation in China
In June this year, the world's first deep learning framework for intelligent interpretation of remote sensing images, LuoJiaNET, and the industry's largest remote sensing image sample library, LuoJiaSET, were launched on the Huawei Shengsi community. This is a collaboration between Academician Gong Jianya, Chief Scientist of Luojia Laboratory, Professor Hu Xiangyun's team, and Huawei's team.
Hu Xiangyun introduced that in the past, the information transmitted by remote sensing satellites had to be manually translated. Faced with massive images, it was urgent to combine artificial intelligence to solve this problem; At the same time, deep learning techniques for remote sensing image processing require large-scale sample libraries as support, and deep learning frameworks and models also need to effectively address the challenges of remote sensing data processing.
"LuoJiaNET is a deep learning framework and toolkit dedicated to intelligent interpretation of remote sensing images. Based on this, anyone can develop various application software to make remote sensing image processing more intelligent." Hu Xiangyun introduced that after collaborating with companies such as Jiahe Technology, LuoJiaNET was successfully applied in the fields of agriculture and natural resources, and quickly implemented and transformed in Hubei, "blooming and fruiting.".
Jiahe Technology Di Shi Tong Smart Agriculture SaaS Service Platform is built based on Shengsi MindSpore and LuoJiaNET, combined with Jiahe's own algorithm library, sample library, and data source library. Previously, data was "translated" manually. Now, after deep learning by computers, the interpretation of agricultural remote sensing information in a township has been shortened from one person per week to human-machine collaboration, which can be completed by one person in two hours. In addition, Wuhan Handari Technology and Shanghai Shuhui have jointly created a natural resource brain using LuoJiaNET, providing intelligent wings for remote sensing monitoring of natural resources such as land, forests, and water.
"Remote sensing is a strategic technology, and LuoJiaNET can achieve autonomous and controllable remote sensing interpretation in China." Hu Xiangyun said that on the LuoJiaNET framework, everyone can contribute data, models, algorithms, and accumulate samples on the LuoJiaSET database, "We hope that they can weave an 'ecological network' and become the most widely used system in China's aerospace information field, especially in the remote sensing industry, within two to three years, and promote it to the world."
It is understood that after more than a year of construction, the Luojia Laboratory focuses on cutting-edge technological innovation research in the field of aerospace information and interdisciplinary research, from 0 to 1. It focuses on four major areas: high-precision spatiotemporal benchmarks and intelligent navigation positioning, aerospace technology key chips and core equipment, aerospace information artificial intelligence methods and security technology, aerospace information detection and real-time intelligent services, promoting institutional and mechanism innovation, and organizing collaborative research, Efforts will be made to break through common key technologies in the field of space technology, carry out technological innovation and industrialization practices, and achieve phased results in satellite and navigation equipment design and development, BeiDou industry technology innovation, intelligent remote sensing open source ecology, cross research between remote sensing and imaging medicine, government consulting services, and other aspects.
