SF-CNN: Signal Filtering Convolutional Neural Network for Precipitation Intensity Estimation.

Chih-Wei Lin,Mengxiang Lin,Xiuping Huang,Sidi Hong

doi:10.3390/s22020551

Chih-Wei Lin, Mengxiang Lin + Show 2 more

Open Access

https://doi.org/10.3390/s22020551

Copy DOI

Abstract

Precipitation intensity estimation is a critical issue in the analysis of weather conditions. Most existing approaches focus on building complex models to extract rain streaks. However, an efficient approach to estimate the precipitation intensity from surveillance cameras is still challenging. This study proposes a convolutional neural network known as the signal filtering convolutional neural network (SF-CNN) to handle precipitation intensity using surveillance-based images. The SF-CNN has two main blocks, the signal filtering block (SF block) and the gradually decreasing dimension block (GDD block), to extract features for the precipitation intensity estimation. The SF block with the filtering operation is constructed in different parts of the SF-CNN to remove the noise from the features containing rain streak information. The GDD block continuously takes the pair of the convolutional operation with the activation function to reduce the dimension of features. Our main contributions are (1) an SF block considering the signal filtering process and effectively removing the useless signals and (2) a procedure of gradually decreasing the dimension of the feature able to learn and reserve the information of features. Experiments on the self-collected dataset, consisting of 9394 raining images with six precipitation intensity levels, demonstrate the proposed approach’s effectiveness against the popular convolutional neural networks. To the best of our knowledge, the self-collected dataset is the largest dataset for monitoring infrared images of precipitation intensity.

Highlights

The understanding of weather conditions has become more critical, and has been discussed for decades due to the dramatic changes in the global climate, in which precipitation intensity is an important issue
The comparison results of the proposed signal filtering convolutional neural network (SF-convolutional neural network (CNN)) with several popular methods were presented
This study demonstrated the performance of the proposed SF-CNN without the proposed components to prove the efficiency of these components

Summary

Introduction

The understanding of weather conditions has become more critical, and has been discussed for decades due to the dramatic changes in the global climate, in which precipitation intensity is an important issue. The estimation of the precipitation intensity is the fundamental technology underlying various applications, for example, farming, weather forecasting, and climate simulation. Various studies have been dedicated to measuring precipitation intensity, and they can be classified into three categories based on the used data sources: gauge-based [1,2], radar-based [3,4], and satellite-based [5,6] approaches. The rain gauge is the earliest and most widely used device to measure precipitation intensity. Various types of rain gauges have been developed in studies to estimate precipitation intensity, but there have been some drawbacks; for example, the rain gauge must be placed on a flat surface perpendicular to the horizontal plane. The rain gauge can only collect the installed local rainfall information

Results

Discussion

Conclusion