The Impact of Noise or Noiseless on Cellular Neural Network to Detect Image Edge

This paper argues that the best effect to remove Gaussian noise is to use wiener filtering, and to remove salt & pepper noise to use median filtering will get a better effect. By using the correlation index, and through the original image adding noise and removing noise, it calculates the correlation index of the removal-noise image of the original image is better than traditional methods which used as average filtering and median filtering and wiener filtering to delete the noise of an image. Specifically, this research paper puts forward two results: one is to provide the add noise image first and then to remove the noise, and then to use CNN to detect the image edge, the other is to provide the noise image first by using CNN to detect edge and then to remove the noise. Via these two results compared with the result of the original image edge detection, the conclusion will be as following: in order to avoiding the impact of noise bring to an image, before the image edge detect, one must deal with the noise first.