Transformation of Engineering Sketch to Valid Solid Object Muhammad Zaini Matondang 1 , Samihah Mardzuki 2 , Habibollah Haron 3 Department of Modeling and Industrial Computing, Faculty of Computer Science and Information Systems, Universiti Teknologi Malaysia, Johor, Malaysia. Email: zaini_edu@yahoo.com 1 amimar84@yahoo.com 2 habib@utm.my 3 Abstract: The evolution of Computer Aided Design (CAD) and its application to solve engineering problem such as interpreting engineering sketch is started with the invention of Sketch Interpreter in 1963 by Sutherland. Since then, previous research on this topics have presented several frameworks to read, analyze, represent, validate, reconstruct, and visualize an object from engineering sketch to a valid solid object. This paper presents a new framework in a complete cycle of transformation object from engineering sketch to valid solid objects. It consists of four phases: the interpreting of engineering sketch, representing the regular line drawing, labeling and validating the line drawing, and reconstructing the validated line drawing. The framework has been tested to few objects and has been implemented by developing a prototype system. The system starts from reading two-dimensional (2D) engineering sketch, thinning the sketch, extracting the feature and represents it using the concept of chain code-scheme. The generated regular line drawing is then labeled using Huffman- Clowes line labeling to identify its validity in representing three-dimensional (3D) object and rejects the impossible drawing. Neural network with back propagation is employed for the subsequently step in deriving the z-values of the visible junctions and the hidden junction of the object. Finally, the reconstructed object is represented as mathematical modeling and visualized. As a conclusion, the combination of thinning algorithm, chain-code scheme, line labeling, neural network, and mathematical modeling in the proposed framework has generated a new invention in the development of sketch interpreter. Keywords: engineering sketch, thinning, chain-code, line-labeling, neural network, mathematical modeling. 1. INTRODUCTION This paper discusses about a framework of a complete cycle of transformation object which started from 2D engineering sketch until a visualization of solid object. There are seven sections in this paper. It is organized as follows; this section discussed some works on sketch interpreter and summarizes them in the next subsection; section two shows the proposed framework; section three until five discuss the detailed of the framework; followed by experimental result in section six and closing with conclusion in the last section. 1.1 Works on sketch interpreter Since 1963, there are lots of researches and reports that focused on sketch interpreter. In this section, few of them are presented. Kanade (1981) demonstrated on how the theory and techniques that exploit assumptions about 3D shapes of objects can be provide a systematic shape-recovery method. Two parts are discussed here. First, the application of Origami-theory that is models the world as a collection of plane surfaces and then recovers the possible shape qualitatively. Second, the techniques of mapping image regularities into shape constraints for recovering the probable shapes quantitatively. Barrow (1981) proposed a computational model that based on constraints on local surface orientation along external and discontinuity boundaries. Techniques for recovering 3D conformation of a space curve and interpolating smooth surfaces from orientation constraints along extreme boundaries are discussed. Marill (1991) emulates model-based human interpretation of line drawing by constructing a program CONSTRUCT. The fundamental principle underlying the program is Minimum Standard Deviation Angle (MSDA) and Necker cube illusion. Wang (1992) proposed an automatic reconstruction of a 3D objects based on constructive solid geometry (CSG) representation. Since the work is based on a CSG representation, it is beyond our discussion. Leclerc (1992) improved Marill’s approach (1991) by improving the objective function that favors both symmetry and planarity. Improving the MSDA function and using a more competent optimization technique reduce time taken to reconstruct. Pugh (1992) developed a program called Viking that acts as interactive sketch interpretation. 2707