Three-dimensional reconstruction of general solid food materials was performed using a reverse engineering method based on a surface cross-sectional design. Digital images of cross-sections of irregular multi-dimensional foodstuffs were acquired using a computer vision system, and image processing was performed to obtain the actual boundaries. These boundaries were then approximated by closed B-spline curves, which were assembled through a lofting technique to construct a geometrical representation of food materials. Considering the reconstructed objects, a procedure based on finite element method was developed to estimate the surface area and volume. The developed finite element method approach was validated against experimental volume values of apples and meat pieces, obtaining an estimation error less than 2%. Surface area prediction equations were proposed from estimated surface area values and weight and volume measurements. Good agreement was found with previously reported results. (C) 2007 Elsevier Ltd. All rights reserved.