Quaternion Convolutional Neural Networks for Heterogeneous Image Processing
This addresses the challenge of efficient learning with small or heterogeneous datasets in image processing, though it appears incremental as it builds on existing quaternion CNN methods.
The paper tackled the problem of color image reconstruction from grayscale training by investigating why quaternion-valued CNNs outperform real-valued CNNs, finding that quaternion convolutional encoder-decoders perfectly reconstructed unseen color images while real-valued ones produced worse, grayscale versions.
Convolutional neural networks (CNN) have recently achieved state-of-the-art results in various applications. In the case of image recognition, an ideal model has to learn independently of the training data, both local dependencies between the three components (R,G,B) of a pixel, and the global relations describing edges or shapes, making it efficient with small or heterogeneous datasets. Quaternion-valued convolutional neural networks (QCNN) solved this problematic by introducing multidimensional algebra to CNN. This paper proposes to explore the fundamental reason of the success of QCNN over CNN, by investigating the impact of the Hamilton product on a color image reconstruction task performed from a gray-scale only training. By learning independently both internal and external relations and with less parameters than real valued convolutional encoder-decoder (CAE), quaternion convolutional encoder-decoders (QCAE) perfectly reconstructed unseen color images while CAE produced worst and gray-scale versions.