Loitering Detection Based on Trajectory Analysis

Abstract

Some potential danger, such as robbery, steal, and drug-dealing activity, can be avoided effectively by detecting loitering behavior in advance. In this paper, we propose a loitering detection algorithm by analyzing the trajectory of pedestrian. Firstly, we propose a trajectory extraction method combined with motion estimation and object detection to adapt to the tracking under a static camera scene. Then we judge the loitering behavior by analyzing the trajectory of object through the duration time and angles variation between the initial point and remaining points on the trajectory. The experiments were performed using several surveillance cameras. And experimental results show that the proposed method provides superior performance in terms of accuracy, robustness, and speed.

Autonomous Abnormal Behaviour Detection Using Trajectory Analysis

“…Centroid points on a trajectory Figure 8 shows the loitering detection algorithm. The method used is adopted from [9]. The two predefined constants used in the beginning of the algorithm are minimum age of the track before its trajectory can be analysed for loitering and the maximum age of the track exceeding which will cause the track to be detected as loitering.…”

“…If the above condition fails then the track's age is compared to maximum track age constant and if it exceeds the constant, the trajectory is considered loitering. Figure 9 shows the trajectory analysis method used adopted from [9]. A point which is outside the trajectory is taken (point O) and A is the initial point while D= {D i | I = 1,2, … m} is a collection of m points with a time interval which is a constant called "angleFrameInteval" (1 second in this research).…”

“…θ i is the angle between the vector ⃗⃗⃗⃗⃗ and the vector ⃗⃗⃗⃗⃗⃗ i . The angle can be calculated in a loop as stated in [9] by using the following formula. When the trajectory is loitering the angle θ i changes periodically.…”

“…When the trajectory is loitering the angle θ i changes periodically. Therefore, the difference between the adjacent angles were calculated next as stated in [9].…”

“…In [9], a loitering an algorithm was proposed to detect loitering. Trajectory extraction was performed and loitering detection was performed by analyzing the trajectory through calculated angles between vectors on the trajectory and a fixed point.…”