|
|
|
| ICA Method Based on 2-D Wavelet Transform and Its Application to Image Separation |
| Wang Ming-xiang; Fang Yong; Hu Hai-ping |
| School of Communication and Information Engineering, Shanghai University, Shanghai 200072, China |
|
|
|
|
Abstract In this paper, a kind of new Independent Component Analysis (ICA) method based on 2-dimensional wavelet transform is proposed. According to the research, the steady-state error of the Natural Gradient Algorithm (NGA) is inverse proportional to the quadratic of the kurtosis of the sources when the probability distribution function of each source is the same. In addition, the kurtosis of the detail coefficients in wavelet domain is always bigger than that of the original images, so the separation precision of ICA method based on 2-dimensional wavelet transform is higher than that of the traditional ICA method. Furthermore, the size of the sub-image in 2-dimensional wavelet domain is a quarter of the source image, so the convergence speed of the proposed method is faster. Finally, this method is used to separate the mixed images. A set of experiments in different situations is done and the simulation results show that the proposed method is effective.
|
|
Received: 23 August 2004
|
|
|
|
|
|
|
|
|
2006, Vol. 28 
