计算机工程与应用 ›› 2016, Vol. 52 ›› Issue (24): 158-161.

• 模式识别与人工智能 • 上一篇    下一篇

基于运动点团的鱼眼图像中多目标检测方法研究

吴健辉1,2,商  橙1,张国云1,2,郭龙源1,2,何  伟1,2,涂  兵1,2   

  1. 1.湖南理工学院 信息与通信工程学院,湖南 岳阳 414006
    2.湖南理工学院 复杂系统优化与控制湖南省普通高等学校重点实验室,湖南 岳阳 414006
  • 出版日期:2016-12-15 发布日期:2016-12-20

Study of moving objects detection in fisheye image based on moving blob method

WU Jianhui1,2, SHANG Cheng1, ZHANG Guoyun1,2, GUO Longyuan1,2, HE Wei1,2, TU Bing1,2   

  1. 1.College of Information and Communication Engineering, Hunan Institute of Science and Technology, Yueyang, Hunan 414006, China
    2.Key Laboratory of Optimization and Control for Complex Systems of Hunan Province, Hunan Institute of Science and Technology, Yueyang, Hunan 414006, China
  • Online:2016-12-15 Published:2016-12-20

摘要: 采用运动点团模式对鱼眼视频序列中的目标检测方法进行了研究和探讨。运动点团模式的运动目标检测分为三个层次,每个层次对应一个具体的检测算法,即基于像素层的背景提取和更新、运动点团层的点团检测和判定及运动目标层的目标标记和跟踪。对三个算法的原理进行了探讨,并结合鱼眼图像的特点进行了算法改进和优化。实验结果表明,以运动点团作为中间检测过程的方法能有效对圆形鱼眼视频序列中的多个运动目标进行检测,特别是图像边缘的大畸变、低分辨率目标,相比传统的检测方法具有更好的检测稳定性和准确性,在大范围智能视频监控中具备很好的实际应用价值。

关键词: 鱼眼图像, 运动点团, 运动目标, 目标检测

Abstract: This paper studies the multi-objects detection in fisheye image that uses the moving blob method. The moving blob detection method is divided into three levels when it uses in moving object detection, and each level corresponds to a specific detection algorithm, which is the background extraction and updating based on the pixel level, the moving blob detection and determining based on the moving blob level, the moving object marking and tracking based on the moving object level. The principle of these three algorithms are discussed, improved and optimized combined with the characteristics of fisheye image. The experimental results show that the objects can be detected effectively and stably in fisheye video sequences using the moving blob method as the intermediate inspection process, especially, when the moving object at the edge of image has the large distortion and low resolution. It has some practical value in wide range of intelligent video surveillance system.

Key words: fisheye image, moving blob, moving object, object detection