A Novel Method for Classification of Ancient Coins Based on Image Textures

A novel approach is proposed to extract textual information for classification and recognition of ancient coins, which is based on tree-structured wavelet transform (TWT) and ant colony optimization (ACO) algorithm. Traditional methods of textural extraction focus on information only in low frequency channels. In this paper, in order to obtain texture information in all channels, incomplete TWT is implemented, and an energy map containing energy value and energy channels is constructed accordingly. Main energy channels and their energy in this map will be recorded as the final textural characteristics for classification and recognition. In image preprocessing, the ACO algorithm combining information entropy is utilized to search an optimal threshold for texture segmentation of ancient coin images. Results of simulation demonstrate special efficiency and feasibility of this novel method.