Air Quality Controlling-Oriented Highly Efficient Method For Monitoring Particulate Matters

Yonghui Zhang,Ke Gu,Zhifang Xia,Junfei Qiao

DOI: https://doi.org/10.23919/CCC50068.2020.9189287

2020-01-01

Abstract:Since Particulate Matters (PMs) are closely related to people's living and health, it has become one of the most important indicator of air quality monitoring around the world. But the existing sensor-based methods for PMs monitoring have remarkable disadvantages, such as low-density monitoring stations and high-requirement monitoring conditions. It is highly desired to devise a method that can obtain the PMs concentration at any location for the following air quality control in time. The prior works indicate that the PMs concentration can be monitored by using ubiquitous photos. To further investigate such issue, we gathered 1,500 photos in big cities to establish a new AQPDCITY dataset. Experiments conducted to check nine state-of-the-art methods on this dataset show that the performance of those above methods perform poorly in the AQPDCITY dataset. To address the above issue, we propose a new photo-based model for PMs monitoring. To be specific, we use the Support Vector Regression (SVR) to incorporate four types of 18 features to obtain a high-density PMs monitoring map. Experiments show that the newly proposed model has achieved superior performance than recently developed methods in the AQPDCITY dataset.