AUGMENTING THE CONCEPT OF SHAPE NUMBERS FOR 3-D SHAPE RECOGNITION

Abstract

Shape recognition plays a crucial role in image processing. This work proposes a method for object recognition, based on boundary detection by combining the concept of Fourier descriptors, wavelet transform and shape number. The main aim is to identify the shape of the object from the features that are drawn out from the boundary of the object. The boundary of the object of interest is extracted and then the shape number is obtained which would serve as the shape descriptor for recognition. Sometimes, the boundary pixels cannot truly represent the actual shape of the object due to the variations introduced during segmentation or edge detection process. So in order to remove the noise components in the boundaries, Fourier series concept is applied to decompose the shape into primary and secondary shapes. In this work, the focus is on primary or global shape. The 2-D shape number for this global shape is found out for 8 connectivity. The same concept is then extended for obtaining the 3-D global shape number to be used as descriptor for shape recognition using 64connectivity.