Quaternion and Split Quaternion Neural Networks for Low-Light Color Image Enhancement
Eduardo De Jesús Dávila-Meza, Eduardo Jose Bayro-Corrochano · IEEE Access · 2023
In this paper, two multilayer quaternion feedforward neural network models are presented. While the first model is based on the quaternion algebra, the second uses thesplit quaternionalgebra. For both quaternion neural networks, a learning algorithm is derived using the extended Kalman filter. In addition, to analyze the performance of these two neural network models, they were applied to recover bright images from obscured images. After the neural network processing, the original scene has a better brightness. The quaternion neural network enhances the color images in the RGB color space, and asplit quaternionneural network uses images in the HSV color space. Therefore, from the results, we can see that the neural network using the HSV color model shows advantages unpublished before, which are not shown by the neural network using the RGB color model. This work shows a novel neural network processing that can be used advantageously for practitioners interested in working using the HSV color model.