Ocean floating creatures in situ imaging have new tools
3 Shenzhen Advanced Hospital team members (Fig. 1) Install underwater floating biomyrograph (Figure 3) on the buoy platform (Fig. 2).
The research team is for map (trainee reporter Yu Wen) Plantami is a key component of marine ecosystems, playing a core role in biological geochemical circulation and carbon cycles, and is also an important foundation for marine fisheries and aquaculture production. Recently, the senior engineer Li Jianping, senior engineer 上海奉贤品茶 in the China Academy of Advanced Technology Research Institute, has developed a new type of underwater imaging instrument system for marine floating creatures in situ monitoring, and is carried out on the large sub-waters. Long sea test.
The latest research results were published in IEEEJOURNALOFOCEANICENGINEERING. It has always been, and the floating biological monitoring depends on artificial network mining and optical microscopy analysis, not only fees, but also faces the dilemma of identification talents, traditional methods cannot meet accurate, timely, continuous and sustainable plankton monitoring needs.
Since the 1990s, there have been many optical in situ imaging techniques for marine floating biological studies. However, the existing underwater floating biomaransphodictors cannot be easily 上海品茶打卡地 integrated into the buoy platform, achieving long-term maritime homework. Therefore, Li Jianping team has developed a buoy to develop a water dark field color imaging system to enhance the long-term, continuous, high frequency, in situ monitoring of marine floating organisms, and make up for the shortcomings of existing observation technology. The imaging system uses a new type of orthogonal layer flash without shadowless reading, not only can achieve high quality underwater real color photography on marine floating biological individuals, but also reduce the leakage of local environment under the water, the largest The degree avoids observation deviation caused by the growth of the floating animals due to the gathering of the float. In addition, the imager also supports different magnification rates, covering a floating biological body length range of 200 microns to 20 mm. In order to reduce the pressure of data storage and transmission, the embedded computing unit equipped with the imager can perform target detection pre-processing in real time after image acquisition, and simply transmit the target image instantly to the cloud server through the wireless network, further identify by the depth learning algorithm of the cloud And quantify to obtain monitoring information for end user remote retrieval. In response to the characteristics of underwater tiny target in situ images, the team developed a target-based image labeling and classification algorithm training strategy to make full use of human intelligence and machine intelligence, implement image label, classifier training, and classification results correction. .
On this basis, the team proposed a dual-volume neurocal-level algorithm that not only efficiently constructs a large-scale image data set containing 90 types of images, but also effectively eliminates the interference of particulate matter on the chopplenization biometric identification in the near-bank water body, and finally realizes High-accuracy fine classification identification of plankton image.
In the 4 years, Li Jianping team integrates the imagery system to the water surface buoy after 4 June 20, 2020, and deployed in the Shenzhen Dasai Bay.
By adopting multiple anti-biological attachment measures, it was successfully recovered on February 25, 2021. In the continuous sea trial for 8 months, the instrument acquires the time series data of the floating biological abundance of the sea area, observed the blood circulation of floating animals, the dynamic changes of the advantageous species, and surprised Dasa Bay for the first time. Sea screw socket outbreak. The outbreak of the disasters and creatures severely threaten the safety of cold source water supply, and the timely and accurate monitoring is an important prerequisite for warning.
The sea trial results prove that the team research and development of marine floating biological observation systems can provide more comprehensive and timely plankton monitoring information, which is expected to become a new tool for marine buoy observations platform. In the future, the team will further study miniaturization, intelligent, networked ocean in situ observation sensors and instruments, which provide new methods and new tools for more efficient, timely and accurate understanding and monitoring marine environments. Editor: Wu Yu 彤].