Parallel approach to expedite morphological feature extraction of remote sensing images for CBIR system

In this paper, we have proposed a parallel approach to the morphological feature extraction process and demonstrated a good computational speedup. Remote sensing images have a typical property of incrementing constantly and each image being very large. Since the images are acquired constantly and hence added into the database regularly in good numbers, hence there is a need to make the feature extraction work more efficient. Moreover morphological features are good texture descriptors and are extremely compute-intensive as well. It is hence attempted to utilize the power of multi-core architecture and expedite the process of feature extraction. These feature descriptors are tested on UC Merced Land Use Land Cover Data set. Experimentation shows that with the use of parallel programming and architecture speed up of as good as 20X is obtained for CCH and RIT feature sets.