A comparative study of region matching based on shape descriptors for coloring hand-drawn animation

The work of coloring hand-drawn animation is done by manually specifying and painting each closed region in line drawings. To make this process more efficient, this research creates associations between closed regions in line drawings of adjacent frames, which need to be colored with the same color. Feature values are first computed from the shape of each closed region, and a cost of associating pairs of closed regions is computed from these feature values. The combination of associations that minimizes the total cost is then computed based on these costs. Three shape descriptors for computing feature values for each closed region are examined: ellipses, Fourier descriptors, and shape context; and the accuracy of making associations using each of them is studied.