A continuing problem with inductive logic programming (ILP) has proved to be difficult to handle. Constraint inductive logic programming (CILP) aims to solve this problem with ILP. We propose a new approach to CILP, and implement a prototype of CILP system called BPU-CILP. In our approach, methods from pattern recognition, such as Fisher's linear discriminant and prototype-based partitional clustering, are introduced to CILP. BPU-CILP can generate various forms of polynomial constraints in multiple dimensions, without additional background knowledge. As a result, the CLP program covering all positive examples and consisting with all negative examples can be automatically derived.