ABSTRACT
As stated in Chapter 3, there are five main shape recognition methods (i.e., statistical approach, syntactic/structural approach, template matching, neural network, hybrid method). However, not all methods are appropriate for each type of shape and application, i.e., the method of choice depends on the properties of the shape to be described and the particular application. The presence of noise can also influence the choice of method. Since the structural approach is based on the local shape features and robust to the noisy condition while template matching can achieve a high recognition rate, a hybrid of structural and template matching can show strong potential for shape matching. Sternby [193] presented a structurally based template matching method, which utilizes the explicit structure of the samples to model the non-linear global variations by a set of affine transformations through a structural reparameterization. Bruneli and Poggio [18] argued that successful object recognition approaches may need to combine aspects of structural feature based approaches with template matching methods. Based on these observations, a hybrid method combining the structural approach and template matching, i.e., the modified Line Segment Hausdorff Distance (LHD), is presented for logo matching in this study.
