Gist, HOG, and DWT-based content-based image retrieval for facial images

Abstract

In today’s world there is a wide range of digitization. Everyone is living in a world of digital through text, images, videos, and many more. This raises need for development of latest technologies for retrieval of digital data of the pictorial form. Content-based image retrieval is an efficient method which automates retrieval of images with respect to its salient features. This paper defines approach to have CBIR on facial images with three different feature extraction methodologies Gist, HOG, and DWT. Each feature extraction method will extract facial features for given query image. These facial features are stored in multidimensional feature vector form then used for classifying given query image using knn algorithm. In this approach, we observe performance of every feature extraction technology with its impact on CBIR using precision and recall values.