基于颜色图直方图的监督颜色恒常性算法

Colour chart histogram based supervise colour constancy

在颜色物体识别等应用中，获取不随光照变化的颜色恒常性特征是问题的关键。监督颜色恒常性方法，即通过在环境中放置校正用的颜色片来计算并消除光照的影响，由于其简单方便而得到了广泛的应用。在本文中，基于对直方图光照不变量的分析，提出了一种新的监督颜色恒常性算法。该算法与传统算法相比较，不需要知道颜色片的反射率和成象系统的通道响应，从而大大简化了实现的要求。文中给出了算法的基本思想，并对实验结果进行了分析。

Calculating illumination invariant attributes is very important in applications such as colour object recognition . Supervised colour constancy,which uses colour chart to calculate and correct illumination changes,is used widely because of its simplicity and efficacy.In this paper,we proposed a new supervised colour constancy algorithm based on histogram analysis of a colour chart.It differs from Novak's method in that it does not need to know the spectral reflectance illumi functions of colour chips and the camera spectral sensitivity,thus simplifies implementation greatly.We give the main thoughts of the algorithm and its implementation in real imaging systems,also with experiment results and analysis.