Managing CAD Data as a Multimedia Data Type Using Digital Watermarking
From Knowledge Intensive CAD to Knowledge Intensive Engineering
Three-dimensional (3D) model, including 3D geometric CAD model, has come to be recognized as a "multimedia" data type, along with such data types as sound, text, still image, and movie. As a data type, each one of these data types has a set of operations associated with it. Representative of operations are input, output, editing, indexing, hyper-linking, compression, similarity-based search, and some form of security, such as authentication, tamper detection, and intellectual property (IP)
... ement. This paper discusses digital watermarking technology for 3D geometric CAD data, which may offer a solution to such security related issues of authentication, tamper detection, and intellectual property management. This paper first examines 3D CAD data as a multimedia data type, followed by discussion on its watermarking opportunities. The paper then presents multiple approaches to watermark mechanical CAD models defined by using parametric curves and surfaces. In particular, two methods that targets NURBS curves and surfaces are presented. The methods, one based on reparameterization and the other based on knot insertion, preserves exact geometric shape of target NURBS objects. 1. As an abstract data type, a media data type is associated with its own set of operations (Figure 1) . A representative set of operations includes input, output, editing, compression, indexing, hyperlinking, similarity (or content) -based search, authentication, and intellectual property (IP) management. The first two of the operations are the minimal necessary for a media data type, but the third, editing, is often necessary as well. The rest of the operations gains importance when the data object is shared, distributed, stored, or broadcast, e.g., via the Internet. In case of the text data type, technologies for many of these operations are fairly well understood and available. Social acceptance of and legal infrastructure necessary for some of these operations, e.g., IP protection, have long been established. For other data types, however, not all of the operations are well understood. For example, in the case of still image data, operations other than input, output, compression, and editing are not well developed. Similarity-based (or, content-based) search and copy protection of still images available on the Internet is still in the research and development stage. Time-dependency that exists in such media data types as audio and movie adds a degree of difficulty. Editing, hyperlinking, and indexing time-varying data embody many issues to be solved. Simply hyperlinking "from this to that" can be quite difficult if "this" and "that" must be specified both in time and space. One wants search for a scene in a movie in which "objects of these kind of shapes/colours/etc. move in this manner". But what is a "scene", and how can we specify objects and their movements?