Online retrieval and indexing of images using multi feature vectors

Abstract

In technology proliferated era of modern world, health care has witnessed huge developments. The cutting edge technologies have paved way for sophisticated and feature rich image processing in medical field using colour tomography and medical resonance imaging. The images obtained using radiological techniques can be stored in a database and the features and implications can be recorded in the database after the analysis of those images by physicians. These databases can be used in obtaining the meaningful analysis of the images obtained through radiology in rural areas of developing countries like India, where sophisticated medical facilities are a dream for many in developing nations. The dataset of images can be divided into training and testing set. Training set of data is utilized to obtain multi feature vectors based on Caffe. Caffe is used in this training with a focus on image recognition. The image feature is a simple image pattern based on which the description of image can be obtained. The features of an image are transformed to a vector space using computer vision algorithms. Moreover a framework has been evolved in this paper to extract the features from image using image descriptors-white box algorithms and neural nets-black box algorithms. We also present the pros and cons of our novel framework for online retrieval and indexing of images using multi feature vectors.