液晶与显示, 2018, 33 (12): 1040, 网络出版: 2019-01-15  

基于改进纹理特征的红外目标跟踪算法

Infrared target tracking based on improved LBP feature
作者单位
河北工业大学 人工智能与数据科学学院, 天津 300130
摘要
针对红外图像背景复杂、杂波干扰严重、相似目标混淆导致的目标跟踪丢失问题, 本文提出了一种改进的低维度纹理特征OCS-LBP(Oriented Center Symmetric Local Binary Patterns, 即方向中心对称的局部二值模式)。首先, 利用此特征可以高效地获取目标图像中每个像素块的梯度方向和幅值信息, 提高了跟踪过程的鲁棒性; 其次, 利用核相关滤波算法结合提取的OCS-LBP特征对目标图像区域进行模型训练; 最后, 根据训练好的模型检测下一帧图像中目标的具体位置。本文在10组红外视频序列上进行了测试, 实验结果表明, 本文算法的精确度和成功率相比于第二名算法分别获得了2.9%和9.9%的提升, 同时在实验设备上算法的平均跟踪速度相比于第二名算法提升了14.15 frame/s。从实验结果可以看出本文提出的算法在红外目标跟踪上表现出较好的鲁棒性、准确性和实时性, 具有一定的研究和实用价值。
Abstract
In order to solve the problem that loss of target caused by complex background, serious clutter interference and similar target confusion in infrared image tracking, this paper proposes an improved low-dimensional texture feature OCS-LBP (Oriented Center Symmetric Local Binary Patterns). First, according to this feature, the gradient direction and amplitude information of each pixel block in the target image can be efficiently acquired, which improves the robustness of the infrared target tracking. Secondly, the kernel correlation filter algorithm combined with the OCS-LBP features is used from infrared target image to training model. Finally, based on the trained model, the target can be detected in the next frame of infrared image. In this paper, the proposed algorithm is tested for 10 video sequences. The result shows that the accuracy and the success rate of the proposed algorithm are improved by 2.9% and 9.9% higher than the second algorithm respectively. At the same time, the average tracking speed of the proposed algorithm on the computer improves 14.15 frame/s. From the experimental results, it can be seen that OCS-LBP feature has better robustness, accuracy and real-time performance in infrared target tracking than traditional features, which has certain research and practical value.

卢杨, 张磊, 郭立媛, 杜若鹏. 基于改进纹理特征的红外目标跟踪算法[J]. 液晶与显示, 2018, 33(12): 1040. LU Yang, ZHANG Lei, GUO Li-yuan, DU Ruo-peng. Infrared target tracking based on improved LBP feature[J]. Chinese Journal of Liquid Crystals and Displays, 2018, 33(12): 1040.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!