Applying an Improved Neural Network to CCD Noise Removal in Digital Images

In this paper, an improved neural network is applied to CCD noise removal in digital image. According to the characteristics of the nonlinear response function in the CCD camera, the denoising scheme is based on the adaptive window sizes and parameter of the filter. The improved NN both approaches to CCD PTC (photon transfer curve) and classifies it in nonlinearity and assigns the optimum coefficient to the filter window and parameter in order to remove the noise appositely. Visual evaluation and method with detailed statistical analysis prove that the filter can remove the noise apparently while reserving the image edge and improving the SNR, the filter also enhances the efficiency and precision of the image processing by improving the deficiency of BP method.