摘要
在低照度环境下,传统的人脸跟踪在跟踪过程中容易受到遮挡、形变等各种因素的影响,所以为实现低照度环境下人脸的精确跟踪,本文提出了改进的HAAR-LBP级联分类器相结合的压缩感知跟踪方法。首先采用HAAR特征进行人脸粗跟踪,然后对得到的目标采用双边滤波进行图像的加强,最后采用LBP特征进行细跟踪。改进的跟踪方法克服了传统方法在低照度情况下跟踪效果不好、容易丢失目标的难点,同时跟踪的精度也得到了提高。通过实验证明在昏暗场景下,改进的算法比传统的算法在准确率方面提高了5%左右,同时实时性也有一定的提高。
In the low illumination environment,the traditional face tracking is easily affected by various factors such as occlusion,deformation and so on.In order to achieve accurate face tracking in low illumination environment,an improved compressed sensing tracking method combining haar-lbp cascade classifier is proposed in this paper.Firstly,the Haar feature is used to track the face,and then the bilateral filter is used to enhance the image.Finally,the LBP feature is used for fine tracking.The improved tracking method overcomes the difficulties of poor tracking effect and easy loss of target in low illumination condition,and the tracking accuracy is also improved.The experimental results show that the accuracy of the improved algorithm is about 5%higher than that of the traditional algorithm,and the real-time performance is also improved.
作者
张林
侯劲
ZHANG Lin;HOU Jing(Sichuan University of Light Chemical Technology,Sichuan 644000,China)
出处
《自动化与仪器仪表》
2020年第11期5-9,共5页
Automation & Instrumentation
基金
国家自然科学基金(No.61902268)
人工智能四川省重点实验室项目(No.2017RZJ02)
四川省科技厅重点项目(No.20ZDYF0919)。
关键词
压缩感知跟踪
级联分类器
双边滤波
低照度
compressed sensing tracking
cascade classifier
bilateral filtering
low illumination
