Comparison of Wavelet-Based Image Processing Methods Before Neural Network Training
Olesya M. Sedykh, Maria Oreshina · 2023
The problem of image classification using neural networks is considered. Various methods of image preprocessing based on wavelet analysis are used to optimize neural network training. Among these methods are the Haar wavelet transformation, the Dobeshi wavelet transformation, as well as combined methods that include wavelet transformations and moments of color.A Python script is developed for the experiments. During the work methods are compared with each other to identify the most effective among others.