Copyright Detection System for Color Based Image Retrieval

Abstract

Nowadays, a logo is the visual representation and forms the foundation of company and organization. Most of the companies want their logos are private and detect copyright rules. They do not want the logos of their companies duplicate. All the logos that have been registered are stored in an image database. This system intends to see if the new logo of an organization is very similar to any of the existing logos in the image database in order to avoid copyright infringements. Among image retrieval methods, color based image retrieval method is applied in this system. This system analyzes and implements three image retrieve methods to extract the desired image. These methods are K-D Tree, Vocabulary Tree and Locality Sensitive Hashing (LSH). Based on implementation and analysis, the system compares running time and accuracy percentage between these three image retrieve methods.