A Novel Track Initiation Method Based on Prior Motion Information and Hough Transform

Jun Liu, Yu Liu, Wei XiongIIP 2016 · DOI

TL;DR

This paper introduces a novel track initiation method for dense clutter environments, combining prior motion information with a modified Hough Transform. By intelligently voting in the parameter space and leveraging target kinematics, the method significantly improves detection probability and reduces false alarms compared to existing techniques, making it more robust and efficient for real-world tracking scenarios.

What Problem Does This Paper Address?

Accurately initiating tracks for multiple targets in environments with a high density of false measurements (clutter) is a fundamental challenge in radar and sensor tracking. Traditional methods struggle with this "dense clutter" problem, leading to either missed targets (low detection probability) or the creation of numerous false tracks that consume valuable resources and complicate subsequent tracking. This is particularly problematic for detecting "dim targets" which have weak signals and are more susceptible to false alarms. The paper aims to develop a more reliable and efficient track initiation method that can distinguish true targets from clutter even in these challenging conditions.

Methods

The core of the proposed method lies in a novel application of the Hough Transform, enhanced with prior motion information. Instead of every measurement voting for all possible lines in the parameter space (as in a standard Hough Transform), this approach intelligently selects which lines to vote for.

Here's a breakdown of the key techniques:

  • Pairwise Measurement Connection: The method connects pairs of observed data points from different time scans. This is a crucial step to avoid losing target information, especially if a target's echo is intermittent.
  • Kinematic Constraint Filtering: Not all pairs of measurements are assumed to be from the same target. A kinematic constraint is applied: the distance between two points must be consistent with a plausible target speed (defined by vmin and vmax). This significantly prunes the number of potential target tracks early on.
  • Parameter Space Mapping: For valid pairs of measurements (z1, z2) from time t1 and t2, the method calculates the corresponding line in the Hough parameter space (p, θ). This involves determining the angle θ and the distance p of the line that passes through these two points.
  • Intelligent Voting: The key innovation is how these parameter space points are used. Instead of a brute-force accumulation, the paper suggests that measurements only vote for a part of the lines passing through them. This is achieved by quantizing the parameter space into cells and accumulating votes. Peaks in this accumulated cell distribution indicate potential target tracks.
  • Prior Motion Information: The kinematic constraints (speed limits) are a form of prior motion information that helps filter out spurious connections. The paper also mentions utilizing "time sequence information and targets' kinematics information" to avoid "single scan data accumulation effect," implying a more sophisticated use of temporal data than simple point-to-point connections.

Datasets

The paper does not explicitly mention using pre-existing datasets. Instead, it relies on simulated data for evaluation. The simulation setup involves a 100,000m x 100,000m surveillance area with 5 targets moving in straight lines at constant velocities. The simulation incorporates realistic radar parameters like PRI, azimuth observation error, and distance observation error. The clutter density (λ) is varied to test the algorithm's performance under different clutter conditions.

Key Insights and Ideas

  • Intelligent Voting is Key: The most significant insight is that not all potential lines need to be voted for. By using kinematic constraints to filter measurement pairs and then intelligently accumulating votes in the parameter space, computational cost and false alarms can be drastically reduced.
  • Robustness to Dense Clutter: The method demonstrates remarkable resilience to high clutter densities, a common failure point for many existing track initiation algorithms.
  • Balancing Detection and False Alarms: The proposed approach effectively balances the need for high detection probability (finding true targets) with the need for low false alarm rates (avoiding spurious tracks).
  • Efficiency Gains: By reducing the search space and the number of votes cast, the method achieves significant improvements in computational cost and initiation speed.

Key Results

The paper presents simulation results comparing its proposed method against several established track initiation algorithms (Heuristic rule, Logic-based, HT, Modified HT).

  • Detection Probability (PD): At a clutter density of λ=50, the proposed method achieves a PD of 89.3%, outperforming others. At λ=100, it maintains a high PD of 63.2%, while other methods drop significantly (e.g., Modified HT at 45.3%, HT at 0.1%).
  • False Alarm Probability (PF): The proposed method shows the lowest PF, particularly at high clutter densities. At λ=100, its PF is 42.1%, significantly lower than the heuristic rule (82.7%), logic-based (76.3%), and HT (95.2%) methods.
  • Computational Cost: Table 2 shows a substantial reduction in computational cost. For λ=100, the proposed method's cost is 8.6x10⁵, orders of magnitude lower than the HT method (3.3x10⁸) and the modified HT method (5.2x10⁷).

Limitations and Future Work

  • Assumptions: The current method assumes targets move in straight lines with constant velocities, especially during the initiation period. This might not hold true for highly maneuverable targets.
  • Parameter Tuning: The performance relies on correctly setting parameters like vmin, vmax, and the threshold N for peak detection in the parameter space.
  • Future Work: The authors suggest that extending the method to handle more complex target maneuvers would be a valuable next step. Further optimization of the parameter space quantization and voting strategy could also lead to even greater efficiency.

Summary updated

BibTeX
@inproceedings{liu2016novel,
  author = {Jun Liu and Yu Liu and Wei Xiong},
  title = {{A Novel Track Initiation Method Based on Prior Motion Information and Hough Transform}},
  booktitle = {IIP},
  year = {2016},
  doi = {10.1007/978-3-319-48390-0\_8},
}
Figure