移动机器人颜色查找表的自动学习方法

利用HSI色彩空间中色度分量对颜色的高区分性，引入颜色查找表的膨胀算法和模糊色的概念，运用EM方法训练高斯混合模型，实现了颜色查找表的自动学习。实验结果表明，该方法有效简化了离线训练的人工干预，消除了样本数量繁多带来的采样和训练时间损耗，运用机器学习的方法提高了颜色查找表的抗干扰性，对于需要快速标记颜色查找表的场合非常有效。

Learning Color Lookup Table Automatically in HSI Color Space

This paper applies the color-discriminability of the hue component in HSI color space and uses EM methods and Gaussian mixture model to learn color lookup table automatically, the dilation algorithm and the softcolor introduced by the paper make the color lookup table more stable and saturated. The results of the experimentation show that the method simplifies the manual intervention of offline-training effectively, reduce the loss of sampling time and training time caused by the numerous samples. Using the machine learning method improves the anti-distribution of the color lookup table and makes the method is useful to classify the color lookup table fleetly.