Since there exist plenty of equipment, such as simple water gauges and monitoring electronic cameras, for the water level measurement of riverways and lakes in the Sanmenxia Reservoir Area of the Yellow River, a water level detection system based on image recognition is designed and developed using existing equipment and its specific process is introduced in this project. Detailed design is conducted for the system using the C++ structured method, and the system function is completed by machine learning and OpenCV technology. The system mainly consists of three parts: water gauge positioning, image preprocessing, and water level data acquisition. Experimental results suggest that the system shows good effect and a high rate for the recognition of water level data; in particular, combined with the characteristic that the water gauge does not move, the system meets the application requirements and has certain practical promotion value.

[1] SHI Hong-hua. Application of Video Image in Reservoir Water Level Detection[D]. Suzhou University, 2017.

[2] LI Yi, LAN Hua-yong, YAN Hua. Research on Water-Level Recognition Based on Image Processing and BP Artificial Neural Network Technology[J]. Yellow River, 2015(12):12-15.

[3] CHENG Gao-qing. Research on Recognition of Water Gauge based on Digital Image Processing[D]. South China University of Technology, 2017.

[4] MAO Xing-yun. Introduction to OpenCV3 Programming[M]. Publishing House of Electronics Industry, 2015.

[5] CHU Ze-fan, ZHANG Zhi-jian, ZONG Ze, et al. Research and Design of Water Level Identification and Monitoring System[J]. Electronic Design Engineering, 2018(12):11-14.

[6] HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]. Proceedings of the IEEE conference on computer vision and pattern recognition, 2016