红外图像的光流计算

图像光流的计算不需要在图像序列中建立特征之间的对应关系，因此光流法在计算机视觉的众多领域，包括运动物体的参数估计和目标跟踪方面都有广泛的应用，由于红外图像的噪声相对较大，光流法很少用于红外图像中目标的运动参数估计和跟踪，这里，使用几种常用的光流计算方法对部分实际红外图像进行了光流场计算。结果表明，当选择合适的方法或计算方法进行一定的改进时，这些红外图像可以得到比较接近实际情况的目标光流场，进而应用于红外图像中的目标分割，运动状态分析与目标跟踪等领域。

Optical flow computation of infrared image

In the process of computing optical flow, the corresponding relationship among features of sequent images isn't required. So the technique is widely used in many fields of computer vision, such as parameter estimation of moving object and target tracking. But it's rarely used in infrared image because of the high noise. Optical flow fields of some real infrared images are computed using several usual methods. The results demonstrate that the optical flow field of target which is close to real situation can be obtained while suitable methods are selected or improved. The optical flow could be used in object segmentation, moving status analysis and target tracking of infrared images.