On Research of Video Stream Detection Algorithm for Ship Waterline
Xing Wang, Hu Chen, QingE Wu, Yinghui Guo · 2020
The detection of the waterline of the ship mainly uses the manual observation of the water gauge scale to calculate the draft depth, which is easily affected by human factors. Therefore, this paper proposes a method based a video stream automatically detect the draft in view of the above problems. Firstly, data reprocessing is performed on each frame image in the video, and images smooth fuzzy is performed to obtain a grayscale image; Secondly, the inflection point operator is proposed to obtain the image edge feature, and the feature extraction kernel is used to detect the waterline. at the same time, the digitized image data is obtained, the water gauge scale value is segmented, and the feature is recognized by the deep neural network method. Finally, the draft is obtained by calculating the difference between the water scale value and the water line pixel. In this paper, several experiments are carried out on the computer to collect the video data. The experimental results show that the proposed kernel model has a high accuracy and strong robustness. At the same time, the interference caused by human factors is removed and the accuracy is improved.