一种基于边界跟踪的算法在自动灯光通信中的应用

Application of an algorithm based on boundary tracking to automatic light communication system

全自动灯光通信的难点是从复杂的图像中正确地识别出信号灯的亮与灭并计算出图像的运动矢量,并以此矢量进行目标跟踪,并且算法复杂度要小以实现实时运算.为此提出了一种改进的边界跟踪算法,该算法通过对边界的简化降低了算法复杂度.实验证明采用该算法的程序在普通PC机上即可实现在复杂环境下对100帧/s、360万像素的高速率图像进行实时处理;再辅以莫斯码序列识别及自适应曝光程序和人机交互手段,能够同时实现信号灯的状态识别与目标跟踪;并且可自动识别对方发送速率,实现最高约80字母/min的通信,在大部分天气环境下通信距离在3.5km以上.

The difficulties of the design of automatic light communication system are to identify the correct signal lights status in the complex environment and to calculate the image motion vectors and then to real- ize the target tracking. An improved boundary tracking algorithm with low complexity is proposed . The experiments show that this algorithm has less time-consuming, and in the complex communication environ- ment it can realize real-time image processing at the speed of 100 frame per second and 3.6 million pixel/ frame. By using this algorithm, the identification of the status of signal light and target tracking are real- ized at the same time and good results are achieved on a normal computer.