%0 Journal Article
%T 基于人工智能与随机场的水位图像识别方法研究<br>Study on Water Level Image Recognition Method Based on Artificial Intelligence and Random Field
%A 武泽恒
%A 陈华
%A 黄凯霖
%A 王俊
%A 郭生练
%A 许崇育
%J Journal of Water Resources Research
%P 445-457
%@ 2166-5982
%D 2022
%I Hans Publishing
%R 10.12677/JWRR.2022.115049
%X 为了提升基于视频图像的自动水位监测的准确度和环境稳定性，提出一种基于人工智能与随机场的水位图像智能识别方法。在Deeplabv3+语义分割模型的基础上引入注意力机制和条件随机场，优化水位线识别，结合摄像头标定结果插值计算水位值。设置3组不同条件的现场水位识别试验，结果表明所提出的改进算法识别结果更精确，具备环境适应性，基本满足水文测验要求。<br />
In order to improve the accuracy and robustness of automatic water level monitoring based on video images, a video water level intelligent identification method based on artificial intelligence and random field is proposed. Based on Deeplabv3+ semantic segmentation model, attention mechanism and conditional random field are introduced to optimize water level recognition, and the water level value is calculated by interpolation combining with camera calibration results. Three groups of water level identification experiments with different conditions are set up. The results show that the proposed improved algorithm is more accurate, has environmental adaptability, and basically meets the requirements of hydrological test.
%K 水位监测，图像识别，语义分割，注意力机制，条件随机场<br>Water Level Monitoring
%K Image Recognition
%K Semantic Segmentation
%K Attention Mechanism
%K Conditional Random Field
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=57293