Implementation of curve fitting algorithm for computer cartoon animation

Abstract

This thesis proposes fast preprocessing and curve fitting methods for the images obtained by scanning the hand-drawings in cartoon animation. The proposed technique can be used as a part of the Digital Inking and Painting system, which is very laborious and tedious procedure in the traditional cell-animation, involving a large number of people. The purpose of this thesis is to provide software environment which allows animators to exert their creativity without limit using computers. The first step in the curve fitting procedure is to perform a preprocessing on the bitmap image to transform it to a convenient form for the subsequent steps. Such processing includes the thresholding to reduce to a binary-image, thinning, and the chain coding. For the thresholding, we find that the Histogram Valley Finding method is the best for a wide range of key drawings. In this thesis, we suggest a new fast thinning method using a 4×4 window which is based on the k×k thinning algorithm. Digitized lines are approximated by using the parametric piecewise-cubic Bezier curves. This thesis also includes an efficient solution to the problem of automatically generating parametric cubic curve approximations to the image obtained from the key drawings in cartoon animation.