Geometric transformations in a lazy functional language

The authors present geometric transformations including translation, scaling and rotation expressed in a lazy functional programming language called Miranda. The language is `lazy´ in the sense that no subexpression is evaluated until its value is required, which is particularly suitable for image processing. The authors demonstrate that, by using this embedded laziness, geometric transformations can be described without inverse mapping, and that the program is natural and modular. In particular, this is more beneficial in consecutive transformations. In addition, they show that higher-order functions also improve modularity of programs, and that, for receiving such benefits, it is important to select appropriate image representations where pixel part and other parameters including origin and length are separated