针对红外目标跟踪过程中目标纹理信息缺乏, 与背景灰度呈现强耦合性, 特别是在遮挡情况下目标特征信息链断裂, 特征信息无法延续的实际跟踪问题, 提出了基于显著特征空间的抗遮挡跟踪算法。首先通过分析红外目标特性, 利用多尺度显著性、对比度和信息熵等信息生成显著特征向量空间, 结合超像素特征距离和空间距离对区域进行聚类融合, 突出目标区域, 生成显著图。然后融合显著区域和原图, 生成多个目标候选区作为跟踪算法输入。最后通过目标的空间分布场矩阵对全局的候选区域进行匹配, 同时建立遮挡检测机制, 基于显著区连通区变化和特征相似度变化曲线对遮挡的起始进行判断, 结合遮挡判定设置模型更新策略。在不同红外测试集上的实验结果表明: 所提算法在遮挡情况下也能达到较好的跟踪效果, 有效增强了跟踪算法的鲁棒性。

Aiming at the lack of target texture information in infrared target tracking, strong coupling between the target and the background, and the case of occlusion causing that the characteristic information cannot be extended, the anti-occlusion algorithm for infrared target tracking with salient feature space was proposed. Firstly, by analyzing the characteristics of the infrared target, the salient feature space was generated by using multi-scale saliency, contrast and information entropy, and the regions were clustered with the super-pixel feature distance and spatial distance to highlight the infrared target area. By quantifying the different regions of the infrared image, the saliency map was generated. Based on the saliency map and the original image, a plurality of target candidate regions was generated as the tracking algorithm′s input. Finally, the spatial distribution field matrices of the infrared target were matched by way of the global candidate regions. At the same time, the occlusion detection mechanism on account of the change of the salience regions and the inter-frame variation curve of feature similarity was established. Experimental results on different IR test sets show that the proposed infrared target algorithm can achieve better tracking performance under the occlusion condition, which effectively enhances the robustness of the tracking algorithm.