基于颜色不变量的多相机目标匹配方法

摘要

针对监控系统中存在复杂光照变化的问题,提出一种基于颜色不变量的多相机目标匹配方法.将人体目标分为头部、躯干和腿部,在导数空间和色度空间中,利用高斯光谱模型提取位于边缘和区域的颜色不变量,通过矩函数得到与平移、旋转和尺度变换无关的矩不变量,用描述子进行目标匹配.实验结果表明,该方法能适应场景中复杂的光照环境变化,以及低分辨率特征不充分等情况,并能获得较高的目标匹配正确率.%According to monitor and control system in light of the existence of complex changes, this paper proposes a multi-camera object matching method based on color invariant. The human target is divided into head, torso and legs. Based on the Gaussian spectrum model, the color invariants of the edge and region are extracted in the derivative space and chromaticity color space respectively. Moment invariants descriptors for object matching are extracted by moment function, which are independent of the translation, rotation and scaling. Experimental results show that this method can effectively resist the complex light changes of the scene and inadequate characteristics under low-resolution conditions, etc., and achieve higher correct rate of object matching.