About the Author

Dr. Marc Ebner has done research in the area of computer

vision since 1996. He has been working in the field of color constancy since the year 2000. Dr. Ebner received the M.S. degree in computer science from New York University, NY, in 1994, the degree Dipl.-Inform. from the Universität Stuttgart, Germany, in 1996 and the degree Dr. rer. nat. from the Universität Tübingen, Germany, in 1999. He received the venia legendi in 2006. At present, he is a lecturer the Eberhard Karls Universität Tübingen, Germany. He has also taught at the University of Würzburg. Courses taught include computer graphics, virtual reality and evolutionary algorithms. He is author of over 50 peer-reviewed publications and is a frequent speaker at international conferences; particularly on areas such as machine intelligence, computer vision, biologically inspired systems, evolutionary algorithms and genetic programming. He serves or has served as a reviewer for many different technical journals such as the Journal of the Optical Society of America A, Image and Vision Computing, Evolutionary Computation, IEEE Transactions on Image Processing, IEEE Transactions on Evolutionary Computation, IEEE Transactions on Systems, Man and Cybernetics - Part B. He is or has been a member of the program committee for many conferences and workshops in the area of evolutionary algorithms. He was co-organizer of the European Conference on Genetic Programming in 2006 and 2007 and also co-organized the EvoStar Conference in 2009 in Tübingen, Germany.

Summary

Light which is reflected from an object varies with the color of the illuminant. However, a sensor which is used inside a digital camera is only able to measure the reflected light. The measured color at  each image pixel varies with the color of the illuminant. A human observer perceives colors as approximately constant. This ability is known as color constancy. Understanding how this may be done is very important for consumer photography. Obtaining color constant descriptors is also very important for automatic object recognition based on color and color image processing in general. In the course of the book, we will first have an in depth look at the human visual system. Next, the reader will learn about the theory of color image formation, color reproduction and different color spaces. In order to obtain a color constant image or approximately color constant image, which does not vary with the illuminant, some assumptions have to be made. A frequent assumption is that the illuminant is constant over the entire scene. We will first discuss algorithms which assume that the illuminant is constant. We then drop this assumptions and discuss algorithms which also work when the illuminant varies within the image. Shadow removal and shadow attenuation are also discussed. Algorithms developed by the author are easy to implement. Similar algorithms could be used by the human visual system. The different algorithms are all provided as pseudo code at the end of the book.

This book is a general introduction into the field of color constancy. You have to have some background knowledge in image processing or computer vision. The book is addressed to professionals working with color images. Students as well as professionals in electrical engineering may consult this book in order to implement color constancy algorithms into scanners, digital cameras or display devices. Researchers who study the human visual system may also find it helpful. Given the algorithmic solutions presented in this book, it may be possible to better understand on how the human visual system processes the available data.

Results Using an Algorithm Described in the Book

Input Image

Output Image

Input Image

Output Image