Image colourisation using linear neighbourhood propagation and weighted smoothing

Image colourisation is the process used to add colour to a grey-level image using computers and is based on the local artificial colours of the image. This study proposes a colourisation method based on linear neighbourhood propagation and weighted smoothing. The method avoids the colour-crossing problem associated with existing methods and enables the achievement of natural and smooth colours. The image is initially coloured by hands, and this is followed by the development of a linear neighbourhood matrix model. The process is completed by global colourisation using the priority method. The final colourised image emerges through a filter-based smoothing treatment from the boundary to the smooth area. Compared to currently popular colourisation methods, the proposed method simplifies the requirements for the initial colourisation and also produces clearer, natural, and higher-quality images.