UAV Visual Surveillance Object Detection and Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing surveillance systems face challenges in detecting and tracking moving objects due to fast motion, irregular camera rotation, low frame rates, and strong illumination changes, especially when using Unmanned Aerial Vehicles (UAVs), which complicates object initialization and registration, particularly for small objects and varying scales.
Innovation Solution
The method involves region-based registration of images, motion segmentation, and hybrid target tracking using Kanade-Lucas-Tomasi feature tracker and meanshift, along with recursive geometric registration and sub-pixel image matching to handle lens distortion and camera motion, enabling accurate detection and tracking of moving objects in UAV surveillance systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If region-based registration is performed on adjacent frames, then registration accuracy is improved, but processing time increases
Solution Approach 1:
The patent divides the image registration process into region-based operations on adjacent frames rather than processing entire images at once. This segmentation approach allows the system to achieve high registration accuracy by focusing on specific regions while reducing overall processing time through parallelization and selective computation.
2Measurement precision
If motion segmentation is performed to detect moving objects, then detection accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent applies motion segmentation by dividing the image into multiple regions and analyzing motion characteristics separately in each region. This allows accurate detection of moving objects while managing computational complexity through localized processing rather than global analysis.
Solution Approach 2:
The patent implements local quality analysis by examining different regions of the image with different processing intensities based on their content characteristics. This enables high detection accuracy in critical regions while reducing computational load in less important areas.
3Measurement precision
If hybrid target tracking using Kanade-Lucas-Tomasi feature tracker and meanshift is applied, then tracking accuracy is improved, but processing time increases
Solution Approach 1:
The patent combines the Kanade-Lucas-Tomasi feature tracker and meanshift algorithms into a hybrid tracking system. This merging of complementary algorithms achieves high tracking accuracy by leveraging the strengths of both methods while managing processing time through coordinated execution and shared computational resources.
4Measurement precision
If recursive geometric registration is performed to handle lens distortion, then registration accuracy is improved, but processing time increases
Solution Approach 1:
The patent applies preliminary geometric transformation models to handle lens distortion before performing the main registration process. This preliminary action reduces the complexity of subsequent recursive registration operations, achieving high accuracy while minimizing processing time by preparing the data in advance.
5Measurement precision
If sub-pixel image matching is performed, then registration accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent applies sub-pixel image matching selectively in critical regions where high precision is most important, rather than uniformly across the entire image. This partial application of excessive action (sub-pixel precision) achieves high registration accuracy where needed while reducing overall computational complexity by limiting the intensive processing to essential areas.
Data Source
AI summary
A system and computer program product for performing visual surveillance of one or more moving objects include registering one or more images captured by one or more cameras, wherein registering the one or more images comprises region-based registration of the one or more images in two or more adjacent frames, performing motion segmentation of the one or more images to detect one or more moving objects and one or more background regions in the one or more images, and tracking the one or more moving objects to facilitate visual surveillance of the one or more moving objects.