基于模糊逻辑的多特征融合的粒子滤波跟踪算法

传统粒子滤波跟踪方法采用单一的特征信息，存在跟踪精度低、鲁棒性弱的缺点，为此本文提出了一种基于模糊逻辑的融合颜色和局部二值模式(LBP)的粒子滤波跟踪方法，提高了目标跟踪的精度。首先，提取出能够表征目标的局部二值模式和基于核函数的加权颜色直方图两种信息；然后，利用模糊逻辑的方法对两种特征进行有效的自适应融合。大量试验表明本文算法在复杂环境下能够进行有效的跟踪，具有较强的鲁棒性和准确性。

Multi-cue Fusion Particle Filter Tracking Algorithm Based on Fuzzy Logic

The tracking results are often disappointed with poor accuracy and robustness because only single cue is adopted in the traditional particle filter algorithm. To solve this problem, a particle filter tracking algorithm based on fuzzy logic fusion of color and LBP is proposed in this paper, which improves the accuracy of target tracking. Firstly, two features, namely, local binary patterns and weighted color histogram features are calculated. Then, the two features are fused adaptively using the fuzzy logic method. Experimental results show that the proposed tracking method has a good tracking performance in a good deal of challenging environment and the results are accurate and robust.