【摘要】 肤色模型的应用大多是建立在小样本训练集上,而目前常用的肤色模型理论却主要是利用大训练集来提高模型的精确性。那么,如何实现一种基于小样本训练集的且具有较高精确度的肤色模型呢?文章在对该领域常用的三种典型肤色模型分析比较的基础上提出了一种多模型相互结合的肤色模型———统计色度空间模型。实验表明在四种模型中效果是最好的。更多还原

【Abstract】 Most skin models are built on small training sets in real applications.To make the model more accurate,enlarging the training set is a main choice.How to build an accurate skin model on a small training set is a hallenging job.This paper proposes a new model-statistical chroma space build on the color histogram and simple chroma-space,also,it analyzes other three popular methods in detail.Experiments show that model performs significantly better than the others.更多还原