Disclosed is an apparatus and method for reconstructing a high-resolution image based on a super-resolution neural network, wherein the apparatus for reconstructing a high-resolution image based on a super-resolution neural network according to an embodiment of the present application includes an input receiving unit for receiving input data to be reconstructed from a high-resolution image, and the Nonlinear mapping is performed by performing an operation on a first layer that calculates input data to extract a patch, a modified activation function that uses a convolution operation result value as an input, or a bidirectional activation function that uses the patch extracted from the first layer as an input. It may include a second layer to be performed and a third layer for restoring a high-resolution image based on a result of a convolution operation using an output value of the second layer as an input.
展开▼