Method for extracting water surface line by utilizing water gauge reflection

Abstract

The invention discloses a method for extracting a water surface line by utilizing a water gauge inverted image, which comprises the following steps of: constructing an identification model and acquiring a video image, identifying the video image through the identification model to obtain a water gauge character, and extracting position information of the water gauge character based on the water gauge character; judging the water gauge characters based on the position information of the water gauge characters to obtain the change trend of the water gauge characters, and grouping the water gauge characters based on the change trend of the water gauge characters to obtain water gauge character groups; and screening the water gauge character set to obtain a character pair, and calculating to obtain the position of the water surface line based on the character pair. The method is simple, the position of the water surface line can be quickly and accurately determined, and the accuracy of identifying the water gauge monitoring water level data based on deep learning is improved.

Description

technical field [0001] The invention relates to the technical field of deep learning, in particular to a water surface line extraction method using the reflection of a water gauge. Background technique [0002] With the development of deep learning and image recognition technology, the method of identifying water gauges based on deep learning to monitor water level data has been widely used in the water conservancy industry. Therefore, the accuracy of the water level monitoring data of this method is seriously affected by the reflection on the water surface. [0003] The difficulty in identifying the water surface in the case of water surface reflection is the determination of the water surface line. The existing methods mainly calculate the position of the dividing line through the light and dark relationship between the water surface and the water surface; use median filtering to eliminate water surface reflection; Modeling, extracting the water gauge area; or designing a...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problem in the above-mentioned prior art that the position of the water surface line cannot be accurately determined and the recognition error of the water surface in the case of the reflection of the water surface, the present invention provides a method for extracting the water surface line using the reflection of the water scale, which extracts the water surface line in the video image based on deep learning technology. The character and position information of the water gauge, using the symmetry of the water gauge reflection, analyzes the relative positional relationship of the characters to determine the position of the water surface line, which can improve the accuracy of water level monitoring data based on deep learning recognition of the water gauge

Images