Waltz Filtering in Java with JSolver
Andy Hon, Wai Chun · 1999
This paper uses the Waltz Filtering Algorithm and the Line-Labelling Problem to illustrate how AI algorithms can easily be implemented using Java and JSolver2 – a constraint-programming class library. The Waltz Filtering Algorithm popularised the technique of constraint propagation – the cornerstone for algorithms to solve constraint-satisfaction problems (CSP). It was initially developed as a computer-vision algorithm to interpret line drawings of threedimensional scenes. This algorithm was originally implemented in MicroPlanner and Lisp. Although most AI algorithms were invented using either Lisp or Prolog, many commercial applications require the use of modern computer languages such as Java. In this paper, we will illustrate how a line-labelling program can developed using constraint-programming techniques provided by JSolver. JSolver implements advanced AI techniques such as constraint propagation, declarative programming, and non-deterministic search in Java. These techniques have been successfully used for scheduling and resource allocation systems.