Practical least-squares for computer graphics

The course presents an overview of the least-squares technique and its variants. A wide range of problems in computer graphics can be solved using the least-squares technique (LS). Many graphics problems can be seen as finding the best set of parameters for a model given some data, and usually these parameters can be estimated using a least-squares approaches. For instance, a surface can be determined using data and smoothness penalties, a trajectory can be predicted using previous information, joint angles can be determined from end effector positions, etc. All these problems and many others can be formulated as minimizing the sum of squares of the errors.