Gate opening ruler reading method

Abstract

The invention discloses a gate opening ruler reading method, relates to the field of intelligent ruler reading, aims to solve the problems in the prior art. The gate opening ruler reading method aims to detect a target set on an opening ruler through a target detection method and obtain the accurate opening of a gate according to the position relationship among targets. According to the invention, after the preset bit set is set, the camera can rotate to each preset bit in the preset bit set and photograph, the target detection algorithm is combined with the digital image processing algorithm, only appropriate target-level manual labeling is needed, and under the condition that the target is not detected by the target detection method, the vernier target is detected again by using a digital image processing algorithm; according to the method, high precision and low omission ratio are ensured under the condition that the real-time requirement is met. Therefore, the analyzable opening degree range is greatly expanded, and the analyzable opening degree range is expanded to ten meters under the condition that the reading precision is guaranteed.

Description

technical field [0001] The invention relates to the field of intelligent ruler reading, in particular to a method for reading a gate opening ruler. Background technique [0002] The gate is used to close and open the control facility of the discharge (discharge) channel. An important part of hydraulic structures, it can be used to intercept water flow, control water level, regulate flow, discharge sediment and floating objects, etc. By effectively adjusting the opening of the gate, it can intercept the water flow, control the water level, and adjust the flow; therefore, it is particularly important to accurately feedback the opening of the gate. [0003] The opening scale is a high-precision measuring instrument used to control the opening of the gate. It has left and right double-row scales, and the left and right scales have their corresponding height ranges (such as the knowledge range of the left scale is 0-4m, and the indication range of the right scale is 3-7m). , me...

AI Technical Summary

Problems solved by technology

[0011] The existing computer vision digital image processing technology still has many problems in the actual implementation of the reading of the opening scale: the lack of robustness, such as reflections, foreign objects, uneven lighting, occlusion, blurring, etc., all of which have greatly affected Precise monitoring and control of gate opening;

[0018] 1. Segment the cursor target through the deep learning semantic segmentation algorithm. This method runs slowly and takes a long time, which does not meet the real-time application requirements; this method requires a large number of pixel-level manual annotations, which is expensive

[0019] 2. Detect the cursor target through the deep learning target detection method, which is not perfect in the case of missed detection and false detection;

[0020] 3. Detect the cursor target through the video motion detection method. This method can only be used in the scene where the camera shoots the video, and the memory and bandwidth required for communication are doubled; this method can only capture the cursor target during the rapid movement of the cursor. However, the reality In the scene, the cursor often stays at the same position for a long time, causing this method to fail to capture the cursor target

Images