Multi-Target Trajectory Detection with Convex Sparsity Priors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing surveillance systems face challenges in detecting and tracking multiple targets in sensor data due to noise interference, computational complexity, and the exponential search space of target trajectories, especially when targets are closely spaced and moving.
Innovation Solution
A method using convex sparsity-inducing formulations for optimization problems, allowing for the detection of multiple targets without pre-enumerated trajectories, by leveraging convex regularization priors and iterative optimization techniques such as group coordinate descent and Viterbi algorithms to efficiently identify target trajectories.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional object detection methods are used, then single target detection is possible, but multiple targets that are closely spaced and interfering with each other cannot be reliably detected
Solution Approach 1:
The patent segments the optimization problem into multiple sub-problems by introducing sparsity priors that separate target detections. The convex optimization is divided into steps where targets are detected and removed iteratively, allowing close targets to be distinguished through sequential processing rather than simultaneous detection.
Solution Approach 2:
The patent introduces sparsity priors as intermediary elements that mediate between the sensor data and target detection. These priors act as a filter that separates true target signals from noise and interfering signals, enabling reliable detection of multiple targets even when they are closely spaced and mutually interfering.
2Measurement precision
If exhaustive search of all possible target trajectories is performed, then complete trajectory identification is achieved, but computational complexity becomes exponential
Solution Approach 1:
The patent applies partial action by searching only through a subset of plausible trajectories rather than all possible trajectories. The sparsity prior restricts the search space to only those trajectories that are supported by the sensor data, eliminating unnecessary computational paths while maintaining detection accuracy.
Solution Approach 2:
The patent changes the optimization parameters by introducing sparsity regularization terms that transform the objective function. This parameter change converts the exponential search problem into a convex optimization problem that can be solved efficiently, reducing computational complexity while maintaining trajectory identification accuracy.
3Productivity
If sparsity priors are introduced to reduce computational complexity, then optimization becomes tractable, but false detections may occur due to noise
Solution Approach 1:
The patent implements feedback through iterative optimization where the sparsity prior is updated based on the optimization results. The algorithm continuously refines the sparsity level and detection thresholds, using feedback from each iteration to adjust subsequent detections, thereby reducing false alarms while maintaining computational efficiency.
Solution Approach 2:
The patent introduces dynamic adjustment of sparsity parameters during the optimization process. Rather than using a fixed sparsity prior, the system adapts the sparsity level dynamically based on the data characteristics and optimization progress, allowing flexible balancing between computational efficiency and detection reliability.
Data Source
AI summary
Provided is a method for detecting the trajectories of one or more targets in the field of view of one or more sensors, the method comprising: receiving one or more sensor frames corresponding to the one or more sensors; defining a space of allowable target states for the one or more sensor frames; specifying a set of potential target trajectories, each comprising one allowable target state for each of the one or more sensor frames; specifying target signal parameters for each of the allowable target states, such that the target signal parameters predict the expected target signal contribution corresponding to the one or more sensor frames; specifying a data fidelity objective to quantify how well the target signal parameters match the one or more sensor frames; specifying a sequence of one or more sparsity objectives to penalize a number of detected targets; determine the trajectories of one or more targets as follows: obtain values for all the target signal parameters in all the sensor frames, the obtained values being initialized values or previously optimized values, for each sparsity objective of the sequence, starting with the obtained target signal parameters, determine new target signal parameters to optimize the sum of the data fidelity objective and the sparsity objective; and storing the final trajectories in memory.