Bibliografie

(eng): A frequently used strategy for road sign classification is based on the normalized cross-correlation similarity to class prototypes followed by the nearest neighbor classifier. Because of the global nature of the cross-correlation similarity, this method suffers from presence of uninformative pixels (caused e.g. by occlusions), and is computationally demanding. In this paper, a novel concept of a trainable similarity measure is introduced which alleviates these shortcomings. The similarity is based on individual matches in a set of local image regions. The set of regions, relevant for a particular similarity assessment, is refined by the training process. It is illustrated on a set of experiments with road sign classification problems that the trainable similarity yields high-performance data representations and classifiers. Apart from a multi-class classification accuracy, also non-sign rejection capability, and computational demands in execution are discussed. It appears that the trainable similarity representation alleviates some difficulties of other algorithms, currently used in road sign classification.

(cze): Návrh klasifikátoru dopravních značek založeného na podobnosti zkoumaného objektu ke třídě značek reprezentované vždy typickou značkou (prototype-based rule). Navržen algoritmus založený na trénování podle množiny prototypů (trainable similarity). Experimenty na několika datových souborech ilustrují vyšší účinnost klasifikátoru v porovnání s klasifikátory až dosud používanými pro klasifikaci značek. Byly testovány i další vlastnosti navrženého klasifikátoru jako robustnost a časová náročnost.