Abstract

                         Corneal reflection extracted from an eye image identifies the relationship between the subject of the image and the scene in front of the subject. The reconstructing scene from corneal reflections provides detailed information about the environment opposite to the subject. It also provides scrutiny about any critical, a subject is encountered with. This research area has significant implications in computer vision, Human-computer interaction, psychology, and image forgery detection.  Digital image processing and computer vision techniques can be used to reconstruct the scene from corneal reflection image, which involves identification of corneal area in the eye, developing eye geometric model to correct the spherical effect of the eye, implementation of super-resolution algorithms to reconstruct the lost visual information present into the environment. 

Introduction

                           We are living in a digital era, where millions of pictures are taken and shared on daily basis. With excessive use of images and advancement in the field of digital image processing, it's not hard to counterfeit any image to create fake scenarios that are referred to as image forgery. The advancements in the field of image processing and computer vision are embraced by many other fields, therefore, these fields deliberately accept their contribution to solve the real time problems.

Eyes are the sensing organ of the human body and have a complex optical system, which accepts the incoming light from the environment and enables humans to perceive the outside world visually. Human eye consists of a transparent layer called cornea on the top center of the eyeball, which acts as a window to the eye. Such as eye anatomy is shown in Figure-1. Light enters to the eyes from the cornea, a large amount of the light refracts and focuses on the retina to develop an image and remaining light reflect back from the cornea to the environment. This reflection ability of cornea makes it able to own front view of an environment on its transparent surface. Eye cornea of a subject in Figure-2 is showing frontal beach’s view on its surface. The use of that corneal reflected image to regenerate the front environmental scene of a subject is referred to as scene reconstruction from corneal imaging. That reconstructed image helps to analyze the environment a subject is in. 

Eyes are the sensing organ of the human body and have a complex optical system, which accepts the incoming light from the environment and enables humans to perceive the outside world visually. Human eye consists of a transparent layer called cornea on the top center of the eyeball, which acts as a window to the eye. Such as eye anatomy is shown in Figure-1. Light enters to the eyes from the cornea, a large amount of the light refracts and focuses on the retina to develop an image and remaining light reflect back from the cornea to the environment. This reflection ability of cornea makes it able to own front view of an environment on its transparent surface. Eye cornea of a subject in Figure-2 is showing frontal beach’s view on its surface. The use of that corneal reflected image to regenerate the front environmental scene of a subject is referred to as scene reconstruction from corneal imaging. That reconstructed image helps to analyze the environment a subject is in.

Reconstruction of a scene from corneal reflection image involves the implementation of image processing techniques. That may involve extraction of cornea part, removal of eyelids, eyelashes and iris texture noise, modeling the geometry of human eye to eliminate the fisheye effect of corneal image and rebuilding the scene by improving the quality of the image.

A reconstructed scene from cornea can be further evaluated by using some important evaluation measures i.e. No-Reference IQA measure(BRISQUE).

Reconstruction of an image by corneal reflection may facilitate a diverse range of fields, i.e. digital image processing, human-computer interaction, computer vision, and most importantly psychology. In the field of human-computer interaction (HCI), human sensing, point‐of‐gaze tracking, and assistance systems can be improved by reconstruction of a corneal image. It may also facilitate the field of computer vision by, image registration, illumination modeling scene reconstruction, and scene/context recognition. Most interestingly, improvements in the field of psychology can be made by corneal scene reconstruction that may help to analyze stimulus-response from the scene.  

There are a lot of challenges encountered, due to the structure of cornea:

•   Size of the cornea is very small as compared to the environment’s scene. Therefore it is hard to find fine details of a large view in the small-sized mirror (cornea).

•   The reflectivity capacity of the cornea is less than 1%. It means cornea does not completely reflect light coming from front scene Therefore a complete scene does not develop on the surface of the cornea. 

•  Iris present in the cornea makes it hard to identify scene properly. Iris has its own texture and color which makes the reflected scene’s visibility low.

•  Human eyes geometry is different. Its cornea has a spherical effect. Therefore there is a need to eliminate fisheye distortion to get environment’s scene properly. 

Along with all above challenges, there are some limitations; to take only that dataset under processing which has a clear image of face or eye, and where there is one to one relationship between eye and environment’s scene. Therefore there must not be anything between cornea and environment that can disturb the complete construction of scene image on the surface of the cornea through reflection.

Team