3D Modified exoskeleton and its applications for shape recognition

This paper presents a new 3D object representation called a modified exoskeleton (mES) which is another type of skeleton that is more stable and less sensitive to rotation and noise than the conventional skeletons. The mES preserves significant characteristics about an object, that are meaningful for object recognition and reconstruction, as the skeleton does. Then a fast and reliable matching algorithm for recognition and classification system which can identify and classify an observed object is introduced. The main advantage of our matching algorithm is that it can recognize the observed 3D object regardless of its orientation and noise. The first experiment was conducted on a set of artificial 3D objects and the average recognition rate of 92.60% based on the mES is achieved. The second experiment was tested on the classification of volumetric lung tumor data sets and the obtained results are promising.