Deep Convolutional Neural Network Algorithm Based on Im2Col and Non-Local Mean Filter

In the big data environment, Deep Convolutional Neural Network is very popular in various fields, but it still has the problems of slow convolution speed and too much noise data. Therefore, how to solve these problems is the direction of our research. We propose a Deep Convolutional Neural Network algorithm based on Im2col and non-local mean Filter (IF-DCNN). The specific method is to use Im2col convolution instead of traditional convolution to accelerate the convolution speed, and use the non-local mean Filter method based on Pearson similarity to remove noise. The experimental results indicate that the proposed method has good performance in Deep Convolutional Neural Network calculation in big data environment, and is suitable for deep convolutional neural network model training of large-scale data sets.