Research On Ice Inversion Identification And Migration Process Based On UAV Low-altitude Remote Sensing

Ice flood was a special hydrological phenomenon in cold region.When ice blocked the river seriously,the impact of ice flood could cause varying degrees of damage to hydraulic structures such as piers and bank revetments,threated the lives and property of people,and further increased the difficulty of ice flood prevention.In river management,it was important to obtain ice change information quickly and accurately during ice flood periods.However,traditional ice monitoring methods were costly and inefficient.To solve the problem,a new ice velocity extraction method based on UAV remote sensing technology was proposed in this article.In this study,choosing the Mohe River section in Heilongjiang as the research object.Based on high-resolution orthoimages in ice flood period,the inversion identification and migration process of the ice concentration and ice velocity were carried out respectively.Moreover,the data were applied to the research on vertical distribution of water velocity under ice.The specific research contents and results were as follows:(1)The advantages and disadvantages of four image threshold-selection methods were compared and analyzed.According to the features of the original ice grayscale images,meanwhile,the ice and background were separated by combining with top-hat transformation and OTSU threshold segmentation.In order to avoid the influence of sunlight and waves,the article adopted area denoising method,and the results showed that the area denoising method was effective and rapid in removing reflective point.A graph showing how ice concentration changed with time had been plotted by extracting the ice concentration based on the ice binary image.OTSU algorithm ran fast and had high segmentation accuracy,which was suitable for the field of ice image segmentation.The top-hat transformation performed brightness equalization processing on the image,and combined with the OTSU algorithm to reduce the error to less than 1.5%.The area denoising method could solve the influence of wave reflection in ice monitoring images and improved the accuracy of image recognition.The maximum ice concentration was 81.05%,and after an ice jam occurred in the upstream,the monitoring results showed that ice concentration dropped rapidly until the ice jam was lifted,the ice concentration returned to 60%.The results were in accord with the situation observed on site.(2)Feature points in drift ice images were then extracted with the SIFT algorithm.Moreover,the extracted feature points were matched with the BF algorithm.According to optimization results obtained with the RANSAC algorithm,the motion trajectories of these feature points were tracked,and an ice velocity field was finally established.The SIFT algorithm provided the advantages of a high precision and notable robustness.Feature points were selected as local extreme points.The color of the river surface was singular.Only ice surfaces,edges and corners could produce dense and stable feature points.The SIFT and RANSAC algorithms were jointly employed to track the feature points,which could realize ice velocity monitoring.The average ice velocities in the research area reached 2.00 and 0.74 m/s,and the maximum ice velocities on the right side of the river center were 2.65 and 1.04 m/s at 16:00 on April 25,2021,and 8:00 on April 26,2021,respectively.The ice velocity decreased from the river center toward the river banks.(3)Building a three-dimensional numerical model of the water tank under different ice concentration.The RNG k-ε turbulence model was used to numerically simulate the water flow under different ice concentration,and verified its correctness through indoor physical experiments.The hydraulic characteristics such as the position of the maximum velocity point,the depth,and the vertical distribution of the velocity were analyzed.Under the same flow rate,the water depth of100% ice-cover flow was more than that of free flow,The average velocity of 100% ice-cover flow was less than the average velocity of free flow,and the maximum velocity of 100% ice-cover flow was less than the maximum velocity of free flow.The position of the maximum velocity point was moved down compared to the free flow.Under the condition of drift ice,due to the increase of local pressure,the maximum velocity of 65% ice-cover flow was more than that of free flow,and the position of the maximum velocity point was between the free flow and the 100% ice-cover flow.The velocity distributed evenly in the core area,which did not follow the logarithmic distribution.Dividing the flow section into the riverbed area and the ice area by taking the maximum velocity area as the boundary,the velocity of each layer followed the logarithmic distribution.The proposed ice information monitoring technique and reported data in this study could provide an effective reference for the prediction of ice flood disasters.