Optimized SIFT: An efficient algorithm for matching medical products images

Nowadays, no one can neglect the effect of smartphone applications in the daily life. Diversity of these applications are enormously increased, from entertainment to education. In the meantime, informative applications have their pros. In this research, producing a drug finder app has been taken into account. Finding adequate information about drugs from their images is a sort of image registration. Automatic registration in medical product image is a problem, which has not been solved well so far. According to the features of medical image nature, the number of SIFT descriptors numerous and this issue would leads to making SIFT or Affine-SIFT (ASIFT) inappropriate for an online matching. Therefore, in this paper, an improvement to the SIFT is presented, which can reduce the processing time. Matching algorithm is adopted to eliminate repeated matching points. An experimental threshold as well as gridding the images are applied for removal the wrong matching points so as to improve the accuracy of matching further. Compared with traditional registration algorithm, the results show that the improved SIFT algorithm has increased both in time-saving and complexity-reducing. Although this algorithm is tested on a drug database with Persian description, conducted experiments on the Google images shows that it is language independent.