Image Colorization Algorithm Based on Graph Signal Processing Using Two-Steps Image Segmentation
Tsukasa Kubota, Kazunori Uruma · 2021
An image colorization algorithm based on a graph signal processing has been proposed, and it achieves a high colorization performance. In order to model the chrominance image, this colorization algorithm constructs the graph based on the image segmentation from the given grayscale image. However, there is room for improvement respect to the image segmentation technique since the size of each object on the image have not been considered. This paper proposes the image colorization algorithm using a spatial region-based image segmentation algorithm based both on QuadTree with nested Multi-type Tree (QTMT) segmentation and Simple Linear Iterative Clustering (SLIC). Numerical experiments, this paper demonstrates the effectiveness of the proposed algorithm by comparing it with previous studies using eight standard images.