A face aging/reverse-aging synthesis method based on face detection and Log-Gabor wavelet is proposed. Source images are first captured by the AdaBoost face detection algorithm and then individually normalized. A reference image from the target age group that closely resembles the test subject’s face is then selected. Because the Log-Gabor wavelet is characterized by a broader bandwidth and takes less time to search a wide range of spectral information then the Gabor wavelet, the Log-Gabor wavelet method is used to determine the aged skin surface topography and the decomposition map is obtained. By adjusting the number of details to be extracted from the decomposition map, we can effectively synthesize facial images for different age groups. Experimental results are verified with wrinkle density estimation.
Journal of Applied Science and Engineering=淡江理工學刊 16(4), pp.375-384