Process Weather Image By Some CNN

摘要

In this paper, a kind of wipe noise according CNN has been put forward. And the different CNN has been used to process weather image. The first stage is that we provide a method to wipe off the noise from image by a kind of CNN (Cell Nerve Network) template, HalfToning Cell Nerve Network (HTCNN). It can wipe off the Gaussian noise and Speckle noise. It's in reason from the simulation. Then we try to process the weather image by HTCNN. Change some parameter to process the weather image. The cloud seems clearer from the processed image. The second stage is that we take a new way to process image to embossment by CNN. We use a the Optimal Edge Detector Cellular Neural Network (OPCNN) to process weather image. Some embossment had appear on the image when it processed by OPCNN. It will be easy to analysis the weather image from the embossment of the picture.