Varying Illumination – Challenging Issue in Face Recognition

Varying Illumination
Three digital images of same person with varying illumination [5]

Face recognition has been used extensively in day to day life since last two decades such as passport processing, revenue, surveillance, face based search, etc. The key issues for face recognition include handling of (i) varying facial pose, (ii) varying illumination, (iii) varying facial expressions, and (iv) the problem of occlusion. This article explain in detail why Varying illumination is a challenging key issue in Face Recognition Applications.

What is Varying Illumination?

Varying Illumination
Figure 1: Images of one person from having different lighting conditions [1].

Normal illumination condition means the light source is in front of face or light source is present at such a location that light on the face is uniformly distributed. That is variation of lighting effect on face is almost zero. The face under investigation is properly visible. The complexity of illumination condition increases as light source direction increases with respect to viewing angle and non-uniformity of illumination on face increases. Such situation is referred to as varying illumination on face.

Complex Varying illumination condition may be to highly non-uniform as shown in figure 1. Original images of one person from extended Yale B face database having different lighting conditions [1] illumination on face i.e. illumination does vary a lot on different part of face image. It is also referred to as highly varying illumination.

Why Varying Illumination a Challenge?

The performance of any face recognition system is adversely affected by facial appearance changes caused by variation in lighting condition [2, 3]. Ambient lighting changes greatly within and between days, and among indoor and outdoor environments.

Varying Illumination
Figure-2: Image two Different persons with different illumination [6]

The variations across illumination in a single face can be very large, while variations between different faces may be small as shown in figure-2 [4, 6]. The same individual, imaged with the same camera and seen with nearly the same facial expression and pose, may appear drastically different with changes in the lighting conditions.

Figure-1 and Figure-2 shows images with different lighting conditions. It is evident that, even though images belong to same person, they appear to belong to different persons. Also, amount of brightness and darkness vary on different regions of faces due to lighting source direction which introduces grayness ambiguity.

All these factors degrade the performance of face recognition system drastically. Hence, varying illumination is a challenging issue in face recognition applications.


  1. K.C. Lee, J. Ho, and D. Kriegman. Acquiring linear subspaces for face recognition under variable lighting. IEEE Trans. Pattern Anal. Mach. Intelligence, 27(5):684–698, 2005.
  2. A. Jain and S. Li. Handbook of Face Recognition. Springer, March 2005.
  3. P. J. Phillips W. Zhao, R. Chellappa and A. Rosenfeld. Face recognition: A literature survey. ACM Computing Surveys, 35(4):399–458, December 2003.
  4. D. Riccio M. Nappi, A. Abate and G. Sabatino. 2D and 3D face recognition: A survey. Pattern Recognition Letters, 28:1885–1906, 2007.
  6. Peter N. Belhumeur, Ongoing Challenges in Face Recognition, Department of Computer Science, Columbia University, New York.