摘要： 提出一种在不同光照和不同背景条件下多个指示色标块检测的算法。在Lab颜色空间的L通道中，在连续帧中根据图像的相似性确定背景图像，当前帧图像与背景图像差分确定运动区域以去除背景干扰和缩小指示色标块查找的区域。在运动区域中进行区域生长，计算运动区域凸包，在凸包中进行采样，并根据权值表赋予权值，通过聚类确定种子，使用基于最小错误率的贝叶斯决策作为生长准则进行生长。实验结果表明，该方法与传统的颜色阈值向量方法相比，应用场景更广，颜色块的检测效果更好。

关键词: Lab颜色空间, 凸包, 贝叶斯决策, 区域生长

Abstract: A color indicator detection algorithm under variant light condition and variant background is presented. Background image is chosen based on the similarity between consecutive frames of a video in L channel of Lab color space. Image differentiation on any frame and the background can identify the motion region, which can reduce the search area for the color indicator. The convex hull of motion region is found and some samples are taken. By assigning the weight, seeds can be determined using clustering method. Region growth is implemented by applying Bayesian decision with minimal error ratio. Experimental result shows that the proposed method is applicable to more conditions and produces compared with traditional color-threshold vector method.

Key words: Lab color space, convex hull, Bayesian decision, region growth

中图分类号: 

• TP391