Color-Based Surface Reflectance Separation for Scene Illumination Estimation and Rendering

Abstract

ACKNOWLEDGMENTSAs humans, at one point in our life, we decide and become determined to achieve certain goals, which for some represent dreams and for others obligations that need to be fulfilled regardless of any kind of motivations. But most of us forget the true meaning of these achievements. We all seek a better living by advancing our knowledge mainly to help our society and facilitate our existence in this world. Knowledge in reality won't match the divine creation of something as small as an ant or as little as a fly, even after combining the knowledge of all humans and other living-beings that God has created throughout time.First and foremost, all praises, acknowledgements, and thanks go to the Lord of the seven heavens and earth, lord of the great and supreme throne, ALLAH (s.w.t.) almighty, above all his creations. He is the one who gave me life, health, strength, wealth, knowledge and Luckily, many people who I was fortunate to meet have contributed to the growth of these seeds in an excellent way. I was very fortunate to meet my advisor, mentor, and major professor Dr. Malek Adjouadi who during the past years has truly contributed to the person I'm today. Not only with his remarkable ways of transferring his knowledge to his students but also by teaching us leadership, motivation, and how to successfully attain and achieve our objectives. The care that he has always shown for our success in addition to his patience and trust has always pushed us to work harder to reach our goals. Exploiting the merits of the MSD method, a scene illumination rendering technique is designed to estimate the relative contributing specular reflectance attributes of a scene image. The image feature subset targeted provides a direct access to the surface ix illumination information, while a newly introduced efficient rendering method reshapes the dynamic range distribution of the specular reflectance components over each image color channel. This image enhancement technique renders the scene illumination reflection effectively without altering the scene's surface diffuse attributes contributing to realistic rendering effects.As an ancillary contribution, an effective color constancy algorithm based on the dichromatic reflectance model was also developed. This algorithm selects image highlights in order to extract the prominent surface reflectance that reproduces the exact illumination chromaticity. This evaluation is presented using a novel voting scheme technique based on histogram analysis.In each of the three main contributions, empirical evaluations were performed on synthetic and real-world image scenes taken from three different color image datasets.The experimental results show over 90% accuracy in illumination estimation contributing to near real world illumination rendering effects.x