Optimization of chamfer masks

This file contains functions to optimize chamfer masks and implement sequential distance transforms.

This submission implements functions to
(1) Optimize chamfer masks using an alternative cost function, called symmetric mean absolute error (SMAPE), which has been widely used in weather forecasting applications
(2) Convert the designed real-valued chamfer masks into integers
(3) Compute distance maps using a sequential chamfering algorithm and the designed chamfer masks
(4) Boost speed of the original chamfering algorithm
**** A paper about this submission has been submitted to the IEEE Transactions of Image Processing for consideration of publication.