A Method of CCD Noise Removal in Digital Images

In this paper, a method for removing noise from digital images corrupted by imaging device is proposed. The CCD noise model is a combination of a mixture of a fixed-pattern noise and multiplicative Gaussian noise and a mixture of signal-independent noise. Based on the model, an adaptive de-noising algorithm to restore images contaminated by the CCD noise sources is developed. Because the great correlation exists among the wavelet coefficients, by doing the multi-level wavelet decomposition to the image and computing the average value of the image´s wavelet coefficients, which can present the singularity of an image, the window size of the adaptive filter can adjust automatically. The scheme will sharpen the edges and smooth the uniform area at the same time. In the end, experiments are done in images with the synthetic noise and the results indicate the de-noising method is efficient.