Skin Research Center, Pierre Fabre Dermo-Cosmétique, Toulouse, France.
IEEE Trans Image Process. 2011 Oct;20(10):2876-85. doi: 10.1109/TIP.2011.2142003. Epub 2011 Apr 15.
Digital imaging devices are increasingly used for color measurements in fields such as forensics, the food industry, and ecological and medical research. Especially concerning the skin, in the follow-up of benign or malignant lesions, it is crucial to ensure that the measured colors are accurate and reproducible. Several color calibration methods having been presented over the last few years. However, the choice of illuminant used remains a major source of color misinterpretation, thus, much effort is being spent in trying to evaluate this a posteriori. The device presented overcomes this problem by integrating its own light source, although corrections in lighting heterogeneity are still required. In this paper, we present a lighting modelling technique used for shading correction which improves color consistency (as assessed by ∆E evaluation versus colorimeter), noise filtering, computation time, and memory consumption for this type of device.
数字成像设备在法医、食品工业、生态和医学研究等领域越来越多地用于颜色测量。特别是在皮肤方面，在对良性或恶性病变进行随访时，确保测量的颜色准确且可重复非常重要。近年来已经提出了几种颜色校准方法。然而，所用光源的选择仍然是颜色误判的主要来源，因此，人们正在努力尝试对此进行事后评估。本文提出的设备通过集成自己的光源克服了这个问题，尽管仍然需要对光照不均匀性进行校正。本文提出了一种用于阴影校正的照明建模技术，该技术提高了颜色一致性（通过与色度计相比评估∆E）、噪声滤波、计算时间和此类设备的内存消耗。
