Research on Dynamic Water Level Recognition of Cabin Based on Improved Retinex Algorithm: Dynamic Water Level Recognition

Abstract

To accurately identify the dynamically changing water level in the ship's cabin, the Retinex algorithm is improved by using the bilateral filtering method firstly, which enhances the cabin image at the edge of the liquid level. then the image is preprocessed by image grayscale, image segmentation, and morphological processing, The PP-YOLO v2 algorithm is used to measure the water level with dynamic characteristics. Finally, the water tank is used to simulate the ship cabin, and the detection results of the algorithm proposed in this paper are compared with the traditional algorithm. The experimental results show that the improved Retinex algorithm combined with the PP-YOLO v2 algorithm has high accuracy in the dynamic water level recognition of the cabin, with a relative error of 0.71%, compared with the traditional algorithm, the combined liquid level recognition accuracy of this algorithm is improved by 2.09%, with strong application value.