Challenges and Applications of Intrinsic Image Decomposition: A Short Review

Diclehan Ulucan, Oguzhan Ulucan, Marc Ebner

Abstract: Intrinsic image decomposition has become an immensely studied problem over the last decades. It holds many challenges but also provides large benefits if it can be solved. In this study, we provide a short review of intrinsic image decomposition algorithms, datasets, and applications, while also addressing the challenges of the field. Aside from creating an algorithm for this under-constrained problem, another challenge is to evaluate the performance of the developed methods since there are certain limitations in existing evaluation strategies. Thereupon, we introduce two new error metrics, namely the ensemble of metrics and the imperceptible weighted score. The ensemble of metrics integrates different perceptual quality metrics in scale-space, while the imperceptible weighted score is the modified version of the well-known \(\Delta E\) metric. We present the usability of our metrics on two datasets by utilizing various intrinsic image decomposition algorithms.