Human emotion recognition to improve driving safety

Abstract

The emotion recognition program is designed to recognize drivers' emotional states by using real-time video. As an important part of the future automobile auto-control system, the emotion recognition program could provide the information about the drivers' emotional state to the system so that it could identify the aggressive behavior of the driver and make necessary correction to avoid accidents. In this emotion recognition program, ELM machine learning algorithm and various image processing operations are used. A specific face detector used in the driving environment and an emotion classifier is developed. The offline testing and the real-time emotion recognition program could effectively detect the face region and classify the emotional state into positive or aggressive. This project is a very important study for the future driving safety improvement.