Single image super-resolution with self-organization neural networks and image laplace gradient operator

Super-Resolution Images Methodology Applied to UAV Datasets to Road Pavement Monitoring

“…The learning-based methods consist of algorithms which can be grouped into external, internal, and convolutional neural networks as main categories, depending on the source of the training dataset [25]. The learning algorithms, on which the external image super-resolution method is based, provide the relationship between low and high-resolution image patches.…”

“…The aforementioned algorithms include several methods of which the convolutional neural network algorithm appears to be the most efficient [26]. This approach is usually performed on images with lots of patterns and textures, but it doesn't work well on the image structures outside the input image and it fails to generate a correct prediction on images of other classes [25,26]. About the deep learning techniques based on neural networks algorithms for image super-resolution, little research is currently available.…”

“…In a recent study, a super-resolution method has been proposed by Ahmadian et al, in which a second-order image gradient allows the edges and details of high-resolution and low-resolution images to be obtained [25]. A competitive learning is employed unlike other neural networks that use the backpropagation method for the weight vectors updating [25].…”

“…In a recent study, a super-resolution method has been proposed by Ahmadian et al, in which a second-order image gradient allows the edges and details of high-resolution and low-resolution images to be obtained [25]. A competitive learning is employed unlike other neural networks that use the backpropagation method for the weight vectors updating [25]. In particular the researchers used old and modern image datasets to train the single image super-resolution algorithm according to the self-organizing neural network, the "k-nearest" neighbour algorithm and Laplace gradient operator, comparing the performance with other previous works and showing that despite the potential of this approach, the processing speed of the algorithm is, however, slower than other traditional methods [25,29,30].…”

High-Precision Target Detection of Remote Sensing Image Based on Feature Enhancement with 6G Technology

“…It supports images with a resolution of 0.72 m and high-resolution video with a resolution of 1.13 m. e Jilin-1 satellite is currently under construction. In January 2018, the number of satellites in orbit reached ten [6]. You et al launched in 2016 the Gaojin-1 series which is my country's first 0.5-meter highprecision remote sensing satellite.…”