Driver’s Fatigueness Estimation using Video Processing in Wavelet Transform Domain

It has been observed that the fatigueless estimation is based on facial geometric features. However, the geometric features are very often prone to vary to a great extent if the facial pose changes slightly. And the results may be completely 1800 out of phase as that of the desired one. The said problem may be thought of analysis of driver’s facial expression in wavelet domain using entropy of low frequency and high frequency sub band of the facial image. Further, the video movie of the driver may be decomposed into images as frames and can be analysed in wavelet domain so as to correlate the facial images to fatigueless level. In the proposed work, following objectives are proposed in order to solve the presented problem: Decomposition of video into 2d frames/images, Wavelet Decomposition of Facial images in LL, LH, HL and HH frequency sub-bands using haar wavelet, Entropy profile of LL and HH sub-band images and Correlation of Entropy to fatigueless Level.