Deep neural network convolutional layer image processing method based on fractional calculus

The invention provides a convolutional layer image processing method based on a fractional calculus deep neural network, and the method comprises the steps: obtaining a to-be-processed image, carrying out the convolution processing, enabling a convolutional layer to comprise cin input channels and coout output channels, enabling each output channel to comprise cin convolution kernels, and enabling the cin input channels and the coout output channels to be communicated with each other; each convolution kernel is defined in the following mode: generating a matrix A with cin rows * cout columns, taking opposite numbers of all elements in the matrix A at intervals of one column to obtain an order matrix A'used for generating the convolution kernel, and generating the convolution kernel according to a fractional calculus calculation formula through each element in the order matrix A '. The feature map output by the constructed fractional order convolution layer simultaneously contains higher-level high-frequency a