基于多特征信息融合粒子滤波的红外目标跟踪

摘要

Since the multi-feature information has perfect detection and adaptation performances,and the particle filtering is an effective method to process the nonlinear and non-Gaussian characteristics of the target tracking model. The advantages of the above items are combined to propose a tracking algorithm based on multi-feature information fusion for the characteristics of the infrared image. The target color and textural feature are adopted in the method to construct the model according to a certain weight coefficient. The model is fused to the frame of particle filtering. The experimental result shows that the tracking algorithm can track the maritime infrared moving target accurately.%多特征信息有较好的检测性能和适应性,而粒子滤波则是一种处理目标跟踪模型的非线性和非高斯特点的有效方法,将两者优点结合并针对红外图像特点,提出一种基于多特征信息融合的跟踪算法,该方法按一定的权值系数利用目标颜色和纹理特征构建模型,并融合于粒子滤波框架中.实验表明该跟踪方法能准确地跟踪海上红外运动目标.