A NEW BAG OF PIXELS APPROACH BASED ON EDGE DETECTION TO FACE IMAGE RETRIEVAL
Mir Hossein Dezfoulian, Mostafa Parchami, Saman Bashbaghi

Abstract
Daily increase in number of digital images over the internet has increased the need of searching among digital image databases. As a special case, Face image retrieval is important in many applications. In this paper, a new method is proposed for human face retrieval. Also this method can be used in face recognition systems. Images in this method are represented as a bag of pixels and for this purpose images are converted to grayscale, then after calculating first order derivation of the grayscale image, Canny edge-detector is applied to highlight face organs like mouth, eyes, eyebrows and ears. Finally pixels of obtained image are clustered into seven clusters. To test the proposed method, CIE Put Image Database is used. Experimental results show that the proposed method has a good performance and high accuracy, such that its recognition rate for face images is 100% over the database.

Full text: PDF