Rockfall hazards pose a significant threat to the safety of personnel and equipment at water conservancy projects

as well as to the overall operational safety of these projects. Therefore

studying efficient and accurate rockfall monitoring methods is crucial for ensuring the safe operation of water conservancy projects in mountainous areas. To address the high labour costs and low efficiency of traditional monitoring methods

this study utilizes various feature extraction algorithms in OpenCV to obtain rockfall feature information

thereby improving rockfall hazard monitoring methods at both the algorithm and hardware levels. At the algorithm level

object detection and tracking algorithms in OpenCV are utilized to detect and track the movement trajectories of falling rocks. By improving the computational speed and memory efficiency of the algorithms

the accuracy and computational efficiency of rockfall hazard identification are enhanced. Based on this

the accuracy of the rockfall identification algorithm based on OpenCV and Kalman filtering is further validated by comparing its performance with that of methods such as the Hough circle method and the optical flow method. At the hardware level

a rockfall early warning system is constructed using professional cameras and single chip microcomputers as core hardware

enabling the real- time transmission

processing

and storage of on- site monitoring images. The aforementioned study demonstrates that the system exhibits excellent feasibility and stability

effectively capturing rockfall trajectories to enable early warning and forecasting

providing more precise information to ensure the safe operation of mountainous infrastructure

including water conservancy projects

 road construction

and tunnel engineering.