基于光流和水平集模型的运动目标自动跟踪方法

Approach on Moving Target Automatic Tracking Based on Optical Flow and Level Set

在分析直升机在天空飞行序列图像特点的基础上,提出了基于光流和水平集模型的运动目标自动跟踪方法。该方法通过对图像进行高通滤波预处理,获得运动目标的特征图像,用特征图像来求解运动目标的光流,用光流对运动区域进行分割,获得运动区域的中心和半径,以该中心和半径的圆作为初始水平集曲线,以光流速度作为曲线演化的外力修改Chan and Vese模型,用改进后的模型对图像进行分割。实验证明,用本文方法能够快速、准确地自动跟踪运动目标。

Through the analysis of the characteristics of the sequence image about the flying helicopter in the sky, an automatic tracking approach on the moving target is proposed based on the optical flow and level set model. The character images are obtained through the image preprocessing by the high pass filter. The optical flow of the moving target is solved by using the character image. The moving area is segmented with the optical flow. Then the center and the radius of the moving area can be obtained. The circularity with the center and the radius is taken as the initial curve of level set. Chan and Vese model is modified by outside force of curve evolution which is obtained by the speed of the optical flow. The images are segmented by using the modified Chan and Vese model. It is demonstrated that the moving target can be tracked automatically, rapidly, accurately with the method.