Paper published in Image and Vision Computing (IMAVIS), Volume 106, 2021