TRADEMARK IMAGE RETRIEVAL USING LOW LEVEL FEATURE EXTRACTION IN CBIR

Trademarks work as significant responsibility in industry and commerce. Trademarks are important component of its industrial property, and violation can have severe penalty. Therefore designing an efficient  trademark  retrieval system  and  its  assessment  for  uniqueness  is  thus  becoming  very  important  task now a days.  Trademark image retrieval system where a new candidate trademark is compared with  already registered trademarks to check  that there is  no  possibility  of  resemblance,  has   been  identified   as  a  major  application  area  of  Content Based Image Retrieval (CBIR).  This paper proposes an automated system for rotation, translation and scale invariant color trademark recognition. Low level feature extraction techniques used in CBIR are proposed for trademark image retrieval. The performance of the trademark image retrieval system is evaluated using precision and recall.