Group-based sparse coding dictionary learning for object recognition

Object recognition is especially challenging when the objects from different categories are visually similar to each other. This paper presents a novel method of group-based sparse coding dictionary learning (GSCDL) to exploit the visual correlation within a group of visually similar object categories for dictionary learning. First, a clustering algorithm is performed to partition the training data into several groups. Then the sparse coding algorithm is utilized to learn an over-complete dictionary for each group. All dictionaries are combined directly to form a global dictionary. A classification scheme is developed to take advantage of the global dictionary that has been trained. The proposed method has been evaluated on popular visual benchmarks. The experiment results show positive effectiveness of the method.