A Video-Based Intelligent Waterlogging Detection Method

Abstract

The invention discloses a video-based intelligent water accumulation detection method. Firstly, Gaussian low-pass filtering is performed on the video frame image, the texture information of the video frame image is extracted, the signal characteristics of the video frame image texture information are analyzed, and the texture information is fused According to the feature, the existence probability of waterlogging can be obtained, so as to realize the preliminary screening of waterlogging areas on the road, reduce the detection space and time range required for the next stage of discrimination, and improve the efficiency of waterlogging discrimination; secondly, in order to improve the positioning accuracy of waterlogging areas, adopt A hybrid deep learning model is used to build a water-logging discrimination network model, which uses surveillance video data of actual scenes as samples to enhance the recognition accuracy of the discriminative network model for typical water-logging areas such as driving trails, wheel splashes, and water ripples. Poisson noise and other methods improve the adaptability of the discriminant network model to noise such as surveillance camera shake, acquisition hardware noise, and complex lighting, and realize the function of accurately locating waterlogged areas on the road.

Description

technical field [0001] The invention relates to using a road monitoring camera to collect video data, and using image processing and deep learning technology to detect and locate waterlogged areas on municipal roads. Background technique [0002] At present, it mainly relies on manual methods to select areas with frequent water accumulation and deploy road water detection systems. This type of system usually includes elements such as water level detection modules, information transmission modules, and monitoring terminals, so as to realize remote monitoring of areas with frequent water accumulation. Road flooding. The advantage of the above-mentioned traditional water accumulation detection method is that it can obtain water accumulation information more accurately and quickly, but the water accumulation area selected based on artificial subjective experience often cannot cope with the dynamic changes of the environment, and needs to be specially constructed for the laying d...

AI Technical Summary

Problems solved by technology

The advantage of the above-mentioned traditional water accumulation detection method is that it can obtain water accumulation information more accurately and quickly, but the water accumulation area selected based on artificial subjective experience often cannot cope with the dynamic changes of the environment, and needs to be specially constructed for the laying detection system Operating conditions and continuous maintenance of the physical equipment at the monitoring location are required. These problems determine that the traditional water accumulation detection method has the disadvantages of high construction and maintenance costs and inflexible use.

Images