Content based image retrieval system based on watershed transform for trademark images

The paper presents the development of content-based image retrieval system to search similar symbols from a collection of trademark images. The system consists of two stages i.e. feature extraction and similarity process. Watershed transform is employed to handle feature extraction process. Two features are derived from this stage i.e. object size and shape from each image. These features are used as the reference value for the next stage which consists of indexing and retrieval process. In this stage, a tree structure is built for handling index assignment that facilitates storing and searching process. Experiment has been conducted to measure system performance and the result shows that the developed CBIR system is useful to retrieve similar symbol from the collection of trademark images.