Fast eyes detection in thermal images
- PDF / 2,579,830 Bytes
- 21 Pages / 439.642 x 666.49 pts Page_size
- 55 Downloads / 195 Views
Fast eyes detection in thermal images Mateusz Knapik1 · Bogusław Cyganek1 Received: 27 August 2019 / Revised: 16 June 2020 / Accepted: 21 July 2020 / © The Author(s) 2020
Abstract In recent years many methods have been proposed for eye detection. In some cases however, such as driver drowsiness detection, lighting conditions are so challenging that only the thermal imaging is a robust alternative to the visible light sensors. However, thermal images suffer from poor contrast and high noise, which arise due to the physical properties of the long waves processing. In this paper we propose an efficient method for eyes detection based on thermal image processing which can be successfully used in challenging environments. Image pre-processing with novel virtual high dynamic range procedure is proposed, which greatly enhances thermal image contrast and allows for more reliable computation of sparse image descriptors. The bag-of-visual-words approach with clustering was selected for final detections. We compare our method with the YOLOv3 deep learning model. Our method attains high accuracy and fast response in real conditions without computational complexity and requirement of a big dataset associated with the deep neural networks. For quantitative analysis a series of thermal video sequences were recorded in which eye locations were manually annotated. Created dataset was made publicly available on our website. Keywords Eye detection · Thermal imaging · Virtual high dynamic range · Drowsiness control sparse descriptor
1 Introduction Although an idea of autonomous cars is taking on popularity, road security still poses a major issue. Systems for self-driving gain new abilities each year. This involves fusing data from multiple sources as well as forces researchers to come up with new and innovative frameworks for data sharing and analysis among connected vehicles, like for example approach presented by Alam et al. in [3]. On the other hand, a key subject in this ecosystem constitute millions of drivers whose state and behavior play an essential role. Drivers Bogusław Cyganek
[email protected] Mateusz Knapik [email protected] 1
AGH University of Science and Technology, Al. Adama Mickiewicza 30, 30-059, Krakow, Poland
Multimedia Tools and Applications
are still required to be fully aware of their environment. Therefore, drivers’ awareness control/monitoring systems can help in this respect [8, 9]. However, to succeed, such systems must be reliable, versatile, and be able to reliably operate in difficult conditions, such as day and night illumination, shocks, temperature, noise, etc., as well as to deal with drivers’ variability in appearance and behavior. On the other hand, such assisting system must provide reliable and continuous real-time information to the higher layers of information processing. Lastly, it cannot be a source of distraction itself. We argue, that thermal imaging offers many benefits. Systems for detecting fatigue driving found much attention in recent times, both from science and the automot
Data Loading...
