一种适应光线变化的运动手势检测与跟踪方法

An Algorithm of Good Illumination Robustness for Detecting and Tracking Moving Gestures

在数字图像处理过程中,运动目标的有效检测是一个重要的前提。文中针对现实环境下光线的多变性,提出了一种有效的运动手势检测和跟踪方法。在三帧差法的基础上,结合Camshift运动目标跟踪法、单高斯建模法和自适应阈值提取方法,从而解决了传统帧差法无法适应光线复杂变化的问题。同时,通过使用Camshift跟踪算法对检测出的手势进行跟踪,并实时建立肤色高斯模型自动适应光线的变化以及其光线变化所带来的干扰。经实验表明,该算法可有效检测出运动目标并进行跟踪,另外还可排除光线发生突变所造成的影响。

In digital image processing, motion target detection is an important premise. An effective moving gesture detection and tracking algorithm is proposed for the variability of light of the actual environment. The three frames difference algorithm is combined with Camshift moving target tracking algorithm, the single Gaussian modeling algorithm and self-adaptive threshold extraction algorithm to solve the problem that traditional frames difference algo- rithms cannot apply to complex changes in light. The Camshift tracking algorithm is used to track the gesture, and a reahime Gaussian model is built to adapt automatically to the changes in light. Experimental results demonstrate that the algorithm can detect and track the moving gesture and avoid the influence of the changes in light effetely.