Fatigue detection based on facial images processed by difference algorithm

Abstract

In this paper a technique for detecting the daily fatigue using the characteristics of local facial image is discussed. To reduce the influence of environment illumination and retain more details of features, the images are preprocessed by the method of color consistent area correction. The texture and color features are derived from the gray level co-occurrence matrix and two kinds of color space, and the reliability and universality of the system is improved by the difference algorithm. Then the error backward propagation algorithm is employed to form a complete detection algorithm. Finally, the method achieves 92.3% accuracy with the self-built image database. With convenient data acquisition source and low computational complexity, the proposed method has a good prospect for the application in the field of home medical devices and mobile health devices.