A template-matching approach to free-form feature recognition

Feature recognition has been an active research area in the past two decades. While many techniques have been developed, most of them are limited to the detection of features that are of regular shapes. Examples include holes, slots, pockets and steps that are commonly used in engineering parts. This paper reports our investigation on a method that is able to detect free-form features. A template matching approach is used where the template represents the shape of the free-from feature to be detected. The feature recognition method works on a two-dimensional parametric space which represents all possible configurations of the template. A match between the template and the object from which the features are to be detected corresponds to a match-curve in the parametric space. Feature recognition is achieved by first searching the rough location of the match-curve followed by searching a precise location of a point on the match-curve. A program has been implemented to verify the feasibility of the approach, and examples are given to demonstrate the major steps in the method. The use of the method in computer graphics applications such as shape blending is demonstrated. The application of the technique in industrial applications such as injection mould design is also discussed