On application of principal component analysis and linear discriminant analysis to control driver's behavior

Abstract

The article is concerned with the problem of realization to control driver's behavior for detection of some tiredness sighs at long-term hard work. The solution of the task is based on the processing images supplied by the DVR (Drive Video Registrator), principal component analysis and linear discriminant analysis. Main attention is paid to the following phases: projection images of face from original attribute area to reduction subarea of principal components, generating the special classifier according to principal components obtained with the linear discriminant analysis. The paper gives the method to increase algorithm efficiency with preliminary processing the Gabor's filter for exclusion face misses due to differences in illumination and another factor. The article estimates the algorithm accuracy and performance.