Papers
Branch Length Similarity Entropy-Based Descriptors for Shape Representation
- Research Fields산업수학기반연구부
- AuthorOhsung Kwon; Sang-Hee Lee*
- JournalJournal of the Korean Physical Society 71 (2017
- Link https://doi.org/10.3938/jkps.71.1075
- Classification of papersSCI
In previous studies, we showed that the branch length similarity (BLS) entropy profile could be successfully used for the shape recognition such as battle tanks, facial expressions, and butterflies. In the present study, we proposed new deors, roundness, symmetry, and surface roughness, for the recognition, which are more accurate and fast in the computation than the previous deors. The roundness represents how closely a shape resembles to a circle, the symmetry characterizes how much one shape is similar with another when the shape is d in flip, and the surface roughness quantifies the degree of vertical deviations of a shape boundary. To evaluate the performance of the deors, we used the database of leaf images with 12 species. Each species consisted of 10 - 20 leaf images and the total number of images were 160. The evaluation showed that the new deors successfully discriminated the leaf species. We believe that the deors can be a useful tool in the field of pattern recognition.
In previous studies, we showed that the branch length similarity (BLS) entropy profile could be successfully used for the shape recognition such as battle tanks, facial expressions, and butterflies. In the present study, we proposed new deors, roundness, symmetry, and surface roughness, for the recognition, which are more accurate and fast in the computation than the previous deors. The roundness represents how closely a shape resembles to a circle, the symmetry characterizes how much one shape is similar with another when the shape is d in flip, and the surface roughness quantifies the degree of vertical deviations of a shape boundary. To evaluate the performance of the deors, we used the database of leaf images with 12 species. Each species consisted of 10 - 20 leaf images and the total number of images were 160. The evaluation showed that the new deors successfully discriminated the leaf species. We believe that the deors can be a useful tool in the field of pattern recognition.