| dc.description.abstract | Computer vision faces challenges when
reconstructing images. in the proposed idea, we tried an
attempt to enhance image quality by providing guidance to a
deep residual convolutional neural network using pre-existing
datasets. Our method leverages an efficient sub-pixel
convolutional neural network (ESPCNN) [1] algorithm to turn
the low-resolution (LR) images into high-resolution (HR) or
super-resolution (SR) images. The ESPCNN algorithm is
designed to enhance and enlarge images, resulting in
high-resolution images that retain the original data. We employ
a convolutional recurrent neural network (CRNN) [2]approach
to reconstruct high-resolution images from under-sampled
data, which can be useful in applications such as medical
imaging, satellite imagery, and surveillance systems. The
method we proposed addresses some of the limitations of
techniques that use convolutional neural networks, such as
FSRCNN[3] and SRCNN[4]. Through research, we determine
the validity of our proposed model. | en_US |
