Image Enhancement for High-Resolution Visual Contents

Abstract

This paper proposes an image enhancement method using gamma neural networks and exponential transformation. When acquiring an image, degradation occurs in very many imaging systems, and the quality of the image acquired by surrounding environmental factors decreases due to the combination of deteriorating elements. Alternatively, work that facilitates post-treatment may be performed by artificially deteriorating for post-treatment directly. However, if the information on these additional tasks is not known, there is a problem that the post-processing process is expensive or additional degradation occurs. To solve this problem, this paper uses a neural network that estimates gamma maps through residual learning for images that require post-processing, and finally applies exponential transformations to perform contrast improvement. The contrast improvement method proposed through the experimental results provides an image with less color distortion compared to the existing method. © 2023 IEEE.