Facial expression invariant person recognition using feature level fusion of visual and thermal images

In this paper feature level fusion scheme of visual and thermal images in wavelet domain is being analysed. Here, Daubechies wavelet transform, termed as db4, coefficients from visual and corresponding coefficients computed in the same manner from thermal images are combined to get fused coefficients. After decomposition up to fifth level (Level 5) fusion of coefficients is done. At the time of generating fused images of coefficients, two different approaches (absolute maximum and absolute minimum) have been followed. Inverse Daubechies wavelet transform of those coefficients gives us fused face images. The main advantage of using wavelet transform is that it is well suited to manage different image resolutions and after decomposition, the high and low frequency sub-band images can be separated. Therefore, the reconstructed fused images have the information of low and high frequencies. These fused images have been passed logarithmic domain (log-ICA) for reduction of dimensions, which is suitable for face images varying with expressions. Finally, those reduced fused images are classified using a multi-layer perceptron. For experiments, IRIS Thermal/Visual Face Database was used. Experimental results show that the performance of the approach presented here gives an average success rate of 94.81%.