Mobile based big data design patent image retrieval system via Lp norm deep learning approach
Jing Su, Bingo Wing‐Kuen Ling, Qingyun Dai, Jun Xiao, Kim Fung Tsang · 2015
This paper proposes a mobile based big data design patent image retrieval system via a deep learning approach. The images are represented via sparse vectors by a dictionary. The joint representation and dictionary design problem is formulated as a mixed L2 and Lp optimization problem. An iterative algorithm is employed for finding a locally optimal solution. Experimental results show that the retrieval accuracy is high.