五、Motion Detection and Estimation（运动检测于估计）

•Motion Detection: Whether image points are moving or not?

•Motion Estimation: How image points move?

Goals of motion detection

•Identify moving objects

•Detection of unusual activity patterns

•Computing trajectories of moving objects

识别移动对象

异常活动模式检测

计算运动物体的轨迹

Applications of motion detection

•Indoor/outdoor security

•Real time crime detection

•Traffic monitoring

室内/室外安保

实时犯罪侦查

交通监控

Many intelligent video analysis systems are based on motion detection.

一、Motion Detection Methods

（1）Hypothesis Testing with a Fixed threshold（固定阀值假设检验）

•How to handle noise:

where W is a spatial windows.

•Compare intensity gradients to handle illumination change:

 

（2）Hypothesis Testing with Adaptive Threshold

•Let Ek be a MRF of all labels assigned at time tk, and let ek be its realization. Based on Bayes criterion, we can write:

•To increase the detection robustness to noise, the temporal differences should be pooled together, for example with in a spatial window Wl centered at l. The hypothesis becomes:

Where N is the number of pixels in Wl.

（3）MAP Detection

•Let

    where q is zero-mean uncorrelated Gaussian noise and

•The overall energy function can be written as:

二、Motion Models

•（1）Spatial Motion Models

•The velocity at position x in the image plane is described by:

•When combined with 3-D affine motion of a planar surface, it leads to:

•（2）Temporal Motion Models

•When the trajectories are linear and the velocity vt(x) is constant between t=tk-1 and τ (τ>t), a linear trajectory can be expressed as:

•A natural extension of the linear model is a quadratic trajectory model:

•（3）Region of Support

•The set of points x to which a spatial and temporal motion model applies is called a region of support, denoted R.

•（4）Observation Models

•Image intensity remains constant along a motion trajectory:

•Using                                                               

•

•Take noise into account:

•Let s be a variable along a motion trajectory. Then:

•Again, when illumination changes, we use the gradients: 

三（plus）、Estimation Criteria

The models discussed have to be incorporated into an estimation criterion that will be subsequently optimized.

•Pixel-Domain Criteria

•Frequency-Domain Criteria

•Regularization

•Bayesian Criteria