Face image synthesis using wavelet transform

The paper proposes a novel method to synthesize a multi-meanings face image, which is produced by two target face images with different interpretations according to distinct viewing distance and visual angles. Firstly, the target face images are decomposed by wavelet transform with the scale that is a function of viewing distance and image height. Secondly, the multi-meanings image viewed in distinct visual angles is reconstructed by elective composition algorithms and variable weight coefficients. Finally, the resultant images are averaged for noise reduction to achieve compelling visual effect. Experimental results show the efficiency and robustness of the proposed method.