{
  "paper_id": "ce60b4558563952bf616ecd949bcacf5",
  "paper_title": "Lightweight People Counting and Localizing for Easily Deployable Indoors WSNs",
  "paper_authors": [
    "Thiago Teixeira",
    "Andreas Savvides"
  ],
  "publication_date": "",
  "publication_venue": "IEEE Journal of Selected Topics in Signal Processing",
  "abstract": "We describe a lightweight method for counting and localizing people using camera sensor networks. The algorithm makes use of a motion histogram to detect people based on motion and size criteria. The motion histogram is an averaged shifted histogram that estimates the distribution of people in a room given the above-threshold pixels in a frame-differenced “motion” image. The algorithm provides good detection rates at low computational complexity. In this paper, we describe the details of our design and experimentally determine suitable parameters for the proposed histogram. The resulting histogram and counting algorithm are implemented and tested on a network of iMote2 sensor nodes. Our implementation on sensor nodes uses a custom sensor board with a commercial off-the-shelf camera, but the motion histogram is designed to easily adapt to ultralow-power address-event motion imagers.",
  "keywords": [
    "Assisted living",
    "human counting",
    "wireless sensor networks"
  ],
  "paper_institutions": [
    "Yale University"
  ],
  "output_language": "zh",
  "provider": "doubao",
  "model": "ark-code-latest",
  "prompt_template": "simple",
  "extracted_at": "2026-08-16T19:47:27.588201Z",
  "template_tag": "simple",
  "summary": "### 摘要\n本文提出了一种轻量级的人员计数与定位方法，用于室内无线传感器网络。该方法基于运动直方图（averaged shifted histogram, ASH）检测人员，通过帧间差分得到运动图像，利用人体尺寸先验构造直方图，并检测直方图峰值以确定人员位置。该方法避免了背景差分中复杂的背景建模和分割问题，计算复杂度低，适合在资源受限的传感器节点上实时运行。我们针对广角镜头和透视畸变设计了三维人体模型投影到图像平面的非均匀分箱，并通过实验确定了最优参数（w=50cm, δ=15cm）。算法在iMote2传感器节点上实现，采用定制摄像头板，运行频率约14Hz。实验结果表明，在1至5人的场景下，正确检测率分别为92.9%、88.6%、80.1%、71.6%和64.3%，大部分误差源于人员进出视野或短暂遮挡。该设计面向低功耗地址事件（AE）运动成像器，便于进一步降低功耗和计算需求，在辅助生活、安防等应用场景中具有易部署、隐私友好等优势。未来可考虑利用视野重叠增强鲁棒性，并改进时间同步协议以减少重复计数。\n\n### 关键要点\n- 本文采用帧间差分而非背景差分，对光照渐变、像素自然波动和静止物体移动等场景具有鲁棒性，避免了复杂背景建模。\n- 运动直方图是一种平均移位直方图（ASH），通过多个不同起始位置的直方图叠加，减少峰位置估计误差，并具有类似低通滤波的效果。\n- 直方图分箱大小w和移位步长δ根据人体平均尺寸设定，实验中w=50cm、δ=15cm为最优参数，可容忍人员坐下或躺下。\n- 针对广角镜头，使用三维人体模型投影到图像平面形成非均匀分箱，提高了图像边缘区域的检测准确性。\n- 直方图填充算法复杂度为O(I_X I_Y (w/δ)^2)，峰值查找为O(H_X H_Y)，整体计算量小，适合传感器节点。\n- 跟踪器采用二分图匹配和匈牙利算法，结合位置和颜色直方图特征（32箱），并引入虚拟峰以保持静止人员的身份。\n- 实验环境为9m×5m房间，天花板高3m，单个162°广角摄像头覆盖约3.2m×2.4m可用区域。\n- 参数实验表明，w在30–50cm、δ在8–15cm时检测错误率最低，小δ易产生假峰，大δ会降低分辨率。\n- 位置精度测试显示，直方图能以100%概率区分相距15cm的两人，在距图像中心1m内（占可用区域66%）性能一致。\n- 网络实现部署6个iMote2节点，非重叠视野，采用星型拓扑，节点本地计数变化时向基站报告。\n- 系统实验结果：1人时正确检测率92.9%，5人时降至64.3%，误差主要源于人员走出视野、遮挡和时间同步偏差。\n- 该方法无需现场标定，只需相机内参近似已知，便于快速扩展节点数量，且可适配超低功耗AE运动成像器。"
}