一种Bessel矩和相位不变性的目标跟踪算法

A way of objective tracking using Bessel moment and phase congruent

为了提高目标跟踪的准确率,提出一种基于Bessel矩、相位不变性和梯度信息的多特征目标跟踪方法。该方法利用Bessel矩形状特征、相位不变性特征和梯度特征,充分发挥Bessel矩高效的形状描述能力,避免相位不变性对图像对比度的影响。与RGB颜色直方图算法和PC-RGB算法相比,本算法可以准确提取目标区域的形状信息和细微的明度变化信息,对目标跟踪具有更强的鲁棒性。实验结果验证了该方法的可行性和有效性。

In order to improve the effectiveness of objective tracking, a new algorithm is proposed based on Bessel moment, phase congruent and gradient magnitude for objective tracking. This method utilizes effective shape descri- bing ability of Bessel moment. The method also can fill up an efficiency of the contrast invariant of phase congru- ent. Compared with the RGB color histogram and PC combined with RBG color histogram, the proposed algorithm can extracts effectively the shape feature and subtle variance of luminance in the target region, which characterizes better and represents more robustly the target. The experiments show feasibility and effect of the proposed method.