基于稠密采样的海上红外目标跟踪算法

Visual Tracking of Infrared Object on the Sea Using Dense Sampling Features

在红外背景下,长时间鲁棒跟踪运动目标是一个具有挑战性的问题.提出了一种基于稠密特征采样(DSIFT)并结合词袋模型与粒子滤波的算法来处理海上红外目标跟踪问题.该算法首先采用DSIFT算法对目标区域及其邻域进行稠密采样并进行特征描述,从而得到包含正负样本的特征向量,然后采用聚类算法构建视觉字典来建立有判别力的目标外观模型.在跟踪过程中,对候选区域稠密采样并用学习得到的视觉字典进行外观表示,然后计算候选区域与目标区域似然,在贝叶斯框架下使用最大后验概率方法实现对目标的准确跟踪.实验结果表明,该算法与相关算法比较,能够有效处理红外海上目标快速运动、外观变化、背景混淆、部分遮挡而导致跟踪性能下降甚至跟踪目标丢失的问题.同时在典型图像序列上,该算法也具有较好的鲁棒性.

Visual tracking of infrared object in complex background on the sea is a challenging problem.In this paper,we propose a novel tracking method based on bag of features and particle filter with dense sampling.Firstly,local image patches within an object region are densely extracted by using sliding windows （DSIFT） and represented as invariant descriptors.Secondly,visual codebooks are generated by clustering algorithms such as K-means.Therefore,the object region is represented as a discriminative appearance model by the vector quantization and the visual codebook.After that,tracking can operate in a Bayesian maximum a posteriori framework.Each candidate region is represented as a distribution of codebook in descriptor space,which then is compared to that of template target model.The experiments have shown the superior performance of our method on infrared object tracking.Moreover,experiments on public benchmark sequences have demonstrated that our method can track object much better than some mainstream algorithms under circumstances of large appearance change,high speed motion,and partial occlusion.