Using Bag of Visual Words and Spatial Pyramid Matching for Object Classification Along with Applications for RIS

摘要

Image analysis, classification and searching has been a perennial frontier in the domain of Computer Vision. Recent developments have shown significant improvement over the aspects of feature extraction, segmentation and image enhancement. With the proliferating rise in computational power, it is both attainable s as well as affordable to perform complex image transforms and computations in hand held devices, thereby making it possible to construct RIS based applications for everyday use. The paper presents a novel approach towards attaining user supplied image searching for online applications such as e-commerce and cloud using the popular Bag of Visual Key points 1 model and integrating spatial pyramid matching 2 . Such models need heavy data mining, strong datasets for training and accurate image feature extrapolation. The said approach aims at yielding a set of matching and homogeneous images for a given user supplied image.