Localization of Eyes Using Haar Cascade Classifiers

Abstract

When it comes to image processing, a system or an algorithm designed to process an image cannot work with an image in its raw format. Image pre-processing plays an important role in transforming or encoding an image in a form that can be comprehended by the algorithm/system. This may include extraction of certain features or combining features to transform the image into a desirable format. One of the most essential pre-processing tasks is the localization of objects/areas of interest in an image. Localization focuses on detecting only the parts of the image that are to be processed while ignoring the rest of the image. This paper describes an effective method in localizing the eye region from an image which contains the upper region of a human face. The dataset used to perform localization of the eye region is the second subset of the CASIA-Iris-Mobile-V1.0. It is done using a combination of various Haar Cascade classifiers.

 Keywords: Localization of Eyes, Digital Image Processing, Image Pre-processing.