RETRACTED ARTICLE: Sub-network modeling and integration for low-light enhancement of aerial images

• Published in: Optical and quantum electronics Vol. 55; no. 11
• Main Authors: Uganya, G., Devi, C. H. Sarada, Chaturvedi, Abhay, Shankar, B. B., Ramesh, Janjhyam Venkata Naga, Kiran, Ajmeera
• Format: Journal Article
• Language: English
• Published: New York Springer US 2023; Springer Nature B.V
• Subjects: Characterization and Evaluation of Materials; Computer Communication Networks; Distortion; Electrical Engineering; Image enhancement; Lasers; Luminous intensity; Machine learning; Optical Devices; Optics; Photonics; Physics; Physics and Astronomy; Pictures; Visual tasks; Aerial images; Nighttime detection; Convolutional neural network; Low-light image enhancement; Image denoising
• ISSN: 0306-8919; 1572-817X
• DOI: 10.1007/s11082-023-05224-7

Poor intensity and contrasts of the pictures produced by picture-acquiring equipment in low-light conditions create a significant barrier to completing other machine-learning activities. This is crucial to advance the study of low-light picture-enhancing techniques to allow the efficient performance of other visual tasks. This research introduces innovative recognition-based neural networks for generating high-quality augmented low-light pictures using raw sensory information to tackle such a challenging task. We initially use an artificial learning approach called CNN (Convolutional Neural networking) to decrease unwanted chromatic distortion and sound. Utilizing the non-local correlations present in the picture, the geographic attention component concentrates on de-noising. A system is directed to improve redundant color characteristics via the channels attention component. In addition, we suggest an innovative pooling level dubbed the reversed shuffle level that picks meaningful data from earlier characteristics in an adaptable manner. Numerous tests show the suggested system’s efficiency in reducing chromatic distortion and disturbance artifacts during improvement, particularly if the original low-light picture contains a lot of disturbance.