Convolution deconvolution neural network method and system

A system for processing an image, comprising:

• a processor;
• and
• a memory storing instructions executable by the processor to perform operations using a convolution-deconvolution neural network, the convolution-deconvolution neural network comprising a first convolution layer, a second convolution layer, and a deconvolution layer coupled to the first convolution layer and the second convolution layer, and the operations comprising:
• filtering by the first convolution layer a first real image to obtain a first feature map therefor with performance of image features improved, wherein the first convolution layer comprises:
• an extracting layer extracting the image feature of the first real image,
• a shrinking layer coupled to the extracting layer and reducing the feature dimension of the extracted feature,
• a mapping layer coupled to the shrinking layer and non-linearly mapping the extracted feature with the reduced feature dimension, and
• an expanding layer coupling the mapping layer with the deconvolution layer and expanding the feature dimension of the mapped feature to obtain the first feature map;
• upscaling by the deconvolution layer the obtained first feature map to improve a resolution thereof, the feature map with improved resolution forming a second feature map;
• and
• constructing, by the second convolution layer from the second feature map, a second real image having enhanced performances and a higher resolution than that of the first real image.

A non-transitory computer readable storage medium for storing computer readable instructions executable by a processor to perform operations using a convolution-deconvolution neural network, the convolution-deconvolution neural network comprising a first convolution layer, a second convolution layer, and a deconvolution layer coupled to the first convolution layer and the second convolution layer, and the operations comprising:

• filtering by the first convolution layer a first real image to obtain a first feature map therefor with performance of image features improved, wherein the first convolution layer comprises:
• an extracting layer extracting the image feature of the first real image,
• a shrinking layer coupled to the extracting layer and reducing the feature dimension of the extracted feature,
• a mapping layer coupled to the shrinking layer and non-linearly mapping the extracted feature with the reduced feature dimension, and
• an expanding layer coupling the mapping layer with the deconvolution layer and expanding the feature dimension of the mapped feature to obtain the first feature map;
• upscaling by the deconvolution layer the obtained first feature map to improve a resolution thereof, the feature map with improved resolution forming a second feature map;
• and
• constructing, by the second convolution layer from the second feature map, a second real image having enhanced performances and a higher resolution than that of the first real image.