WOLFRAM

Wolfram 语言与系统参考资料中心

ImageFeatureTrack[{image1,image2,…,imagen}]

跟踪图像从 image1 到 imagen 的特征.

ImageFeatureTrack[video]

跟踪 video 各帧中的特征.

ImageFeatureTrack[input,pts]

从第一个图像或视频帧中的初始点集 pts 开始跟踪特征.

更多信息和选项
Details and Options Details and Options
范例  
基本范例  
范围  
选项  
Masking  
MaxFeatures  
应用  
互动范例  
巧妙范例  
参见
相关指南
历史
按以下格式引用：

ImageFeatureTrack

ImageFeatureTrack[{image1,image2,…,imagen}]

跟踪图像从 image1 到 imagen 的特征.

ImageFeatureTrack[video]

跟踪 video 各帧中的特征.

ImageFeatureTrack[input,pts]

从第一个图像或视频帧中的初始点集 pts 开始跟踪特征.

更多信息和选项

  • ImageFeatureTrack 也称为特征跟踪，随着时间的推移跟踪特征点.
  • ImageFeatureTrack 可处理 Video 对象、AnimatedImage 对象或图像序列.
  • ImageFeatureTrack[input] 自动确定第一幅图像或第一帧视频中待追踪的特征点集.
  • 在追踪图像序列或 AnimatedImage 中的点时，会返回一个列表 {pts1,pts2,…,ptsn}，其中 ptsi 表示特征在imagei中的位置.
  • 坐标 {x,y} 被认为是在标准图像坐标系中.
  • 如果在某些图像或视频帧中无法找到点，所有后续帧坐标返回 Missing[].
  • 可以指定下列选项：
  • Masking All感兴趣的区域
    MaxFeatureDisplacement15相邻图像对应点之间的最大距离
    MaxFeatures 100跟踪点的最大数目
    MaxIterations20最大迭代次数
    Tolerance0.03匹配点的容差
  • 设置 MaxFeatureDisplacement->{r1,r2} 下，可以指定竖直和水平方向范围不同的搜索窗口.
  • 设置 Masking->roi 下，点集被限制，以满足 image1 所返回的点 pts1 都落在感兴趣的区域 roi 内.
  • 后续图像之间的跟踪根据迭代算法计算. 设置 MaxIterations->n 和 Tolerance->tol 下，算法在迭代替换少于 tol 或者迭代次数超过 n 时，算法终止.

范例

打开所有单元 关闭所有单元

基本范例  (1)

跟踪从一个图像到另一个的指定点：

Wolfram Language code: ImageFeatureTrack[{[image], [image]}, {{141, 86}, {98, 83}}]

范围  (4)

自动找到要跟踪的初始点集：

Wolfram Language code: ImageFeatureTrack[{[image], [image]}]

可视化被跟踪的点：

Wolfram Language code: images = {[image], [image]}; pts = ImageFeatureTrack[images, MaxFeatures -> 15]; MapThread[HighlightImage, {images, pts}]

自动跟踪视频中的点：

Wolfram Language code: pts = ImageFeatureTrack[v, MaxFeatures -> 15];
Wolfram Language code: HighlightImage[IconizedObject[«frame»], {RandomColor[Hue[_]], Line[DeleteMissing[#]]}& /@ Transpose[pts], {...}]

跟踪视频中指定的点：

Wolfram Language code: v = \!\(\*VideoBox["![Embedded Video Player](video://content-68hlk)"]\);
Wolfram Language code: pts = ImageFeatureTrack[v, {{365, 85}, {295, 155}}];
Wolfram Language code: HighlightImage[IconizedObject[«frame»], {RandomColor[Hue[_]], Line[DeleteMissing[#]]}& /@ Transpose[pts], {...}]

选项  (2)

Masking  (1)

仅仅跟踪指定遮罩中的点：

Wolfram Language code: images = {[image], [image]};mask = [image]; pts = ImageFeatureTrack[images, Masking -> mask, MaxFeatures -> 15]; MapThread[HighlightImage, {images, pts}]

MaxFeatures  (1)

默认情况下，返回所有跟踪的感兴趣的点：

Wolfram Language code: ImageFeatureTrack[{[image], [image]}]//Round//Transpose//Tabular

指定需要跟踪的最大点数：

Wolfram Language code: ImageFeatureTrack[{[image], [image]}, MaxFeatures -> 5]//Round//Transpose//Tabular

应用  (3)

显示两个图像中的点的光流：

Wolfram Language code: pts = ImageFeatureTrack[{im1 = [image], im2 = [image]}, MaxFeatures -> 20]; Show[ImageAssemble[{im1, im2}], Graphics[{Green, PointSize[.02], MapThread[If[#2 === Missing[], {Cyan, Point[#1]}, Arrow[{#1, #2 + {ImageDimensions[im1][[1]], 0}}]]&, pts]}]]

跟踪图像之间的图像拐角：

Wolfram Language code: im1 = [image];im2 = [image]; corners = ImageCorners[im1, 5, Automatic, 15]
Wolfram Language code: ImageFeatureTrack[{im1, im2}, corners]

追踪舞台上跳舞女子面部的某一点，以分析舞蹈动作：

Wolfram Language code: video = \!\(\*VideoBox["![Embedded Video Player](video://content-x2tv8)"]\);
Wolfram Language code: points = ImageFeatureTrack[video, {{366, 231}}]; HighlightVideo[video, points]

表演者敲击的次数可以根据追踪点的 坐标计算得出：

Wolfram Language code: heights = points[[All, 1, 2]]; ListLinePlot[heights]

求峰值：

Wolfram Language code: peaks = FindPeaks[heights]; ListLinePlot[{heights, peaks}, IconizedObject[«opts»]]

根据峰值数量计算跳跃/点击的频率：

Wolfram Language code: Quantity[Length[peaks], "Steps"] / Duration[video]

互动范例  (1)

跟踪一系列图像中的点：

Wolfram Language code: i = [image]; imgs = ImageRotate[i, # Degree, Full, Background -> 1]& /@ Range[0, 90, 5]; pts = ImageFeatureTrack[imgs, MaxFeatures -> 7]; ListAnimate[MapThread[HighlightImage, {imgs, pts}]]

绘制每个跟踪特征的运动路线：

Wolfram Language code: ListLinePlot[Transpose@pts]

巧妙范例  (1)

跟踪粒子：

Wolfram Language code: images = {[image], [image]}; {w, h} = ImageDimensions[images[[1]]]; res = ImageFeatureTrack[images, Flatten[Table[{x, y}, {x, .5, w - .5, 20}, {y, .5, h - .5, 20}], 1]];
Wolfram Language code: Graphics[{Red, If[FreeQ[#, _Missing], Arrow[{#}]]& /@ Transpose[res]}]
Wolfram Research (2012)，ImageFeatureTrack，Wolfram 语言函数，https://reference.wolfram.com/language/ref/ImageFeatureTrack.html （更新于 2025 年）.

文本

Wolfram Research (2012)，ImageFeatureTrack，Wolfram 语言函数，https://reference.wolfram.com/language/ref/ImageFeatureTrack.html （更新于 2025 年）.

CMS

Wolfram 语言. 2012. "ImageFeatureTrack." Wolfram 语言与系统参考资料中心. Wolfram Research. 最新版本 2025. https://reference.wolfram.com/language/ref/ImageFeatureTrack.html.

APA

Wolfram 语言. (2012). ImageFeatureTrack. Wolfram 语言与系统参考资料中心. 追溯自 https://reference.wolfram.com/language/ref/ImageFeatureTrack.html 年

BibTeX

@misc{reference.wolfram_2026_imagefeaturetrack, author="Wolfram Research", title="{ImageFeatureTrack}", year="2025", howpublished="\url{https://reference.wolfram.com/language/ref/ImageFeatureTrack.html}", note=[Accessed: 14-November-2026]}

BibLaTeX

@online{reference.wolfram_2026_imagefeaturetrack, organization={Wolfram Research}, title={ImageFeatureTrack}, year={2025}, url={https://reference.wolfram.com/language/ref/ImageFeatureTrack.html}, note=[Accessed: 14-November-2026]}