Efficient trademark retrieval using weighted image features: A GPU based approach - Brossura

Sultana, Madeena; Shorif Uddin, Mohammad

Brossura

ISBN 10: 3639375696 ISBN 13: 9783639375695

Casa editrice: VDM Verlag Dr. Müller, 2011

Riassunto

Large image databases find diverse applications in real-life situations. It is essential to develop an efficient technique to grasp required information from these databases. Fast retrieval and robustness are two main criteria for efficient recognition of an object using image databases. Keeping this in mind, the present research work describes a hybrid method using a weighted combination of multiple image features mainly color histogram, distance, and moment parameters for trademark recognition. The computational power is accelerated by employing a low cost NVIDIA's Compute Unified Device Architecture (CUDA) enabled GPU. Our experimental results show that this new technique provides more robust performance than either of the individual methods and our GPU implementation requires less than 50% of the CPU computation time. This experimentation demonstrates that the method based on multiple weighted images features with cooperative implementation between the CPU and GPU is an effective way of image retrieval from large databases.

Su questo libro

L'autore

Madeena Sultana: Received her M.S. and B.Sc. in Computer Science and Engineering(CSE) from Jahangirnagar University, Bangladesh in 2011 and 2008, respectively. She is a faculty member at the Dept. of CSE in the University of Liberal Arts Bangladesh. Mohammad Shorif Uddin, PhD: Professor, Dept. of CSE, Jahangirnagar University, Bangladesh.

Dettagli dell?edizione

• Editore: VDM Verlag Dr. Müller
• Data di pubblicazione: 2011
• Lingua: Inglese
• ISBN 10: 3639375696
• ISBN 13: 9783639375695
• Rilegatura: Copertina flessibile
• Numero di pagine: 76
• Contatto del produttore: non disponibile
• Persona responsabile: non disponibile