Abstract:Long-term visual target tracking of unmanned aerial vehicles (UAVs) is a challenging and basic research topic. In recent years, many visual object tracking methods have been proposed based on the kernel correlation filtering algorithm and achieved good results. These algorithms have good performance in short-term tracking, but when the target is occluded or disappears from view, the original update strategy may lead to tracker drift. Based on these issues, this paper proposes a long-term kernel correlation filtering and speeded-up robust features (KCFSURF) target tracking algorithm for UAVs in the process of long-term target tracking due to target occlusion or loss. The algorithm takes the KCF target tracking algorithm as the framework, introduces the strategy of searching and locating the target after occlusion or loss, and uses the peak side lobe (PSR) ratio to determine whether the target is covered, blocked, or lost. When the target is occluded or lost, the SURF-random sample consensus (RANSAC) target retrieval matching strategy is introduced to rematch the target and select the box. The new samples are input into the KCF algorithm to continue tracking the target. To verify the superiority and feasibility of the proposed algorithm, the OTB100, UAV123, and Temple-color-128 dataset is selected to evaluate and analyze the algorithm quantitatively and qualitatively. The evaluation results show that KCFSURF can rediscover the target after it is blocked or lost, realizing long-term stable target tracking. Finally, the effectiveness of the KCFSURF algorithm is verified in an S500 UAV target tracking scene.

Airport Pedestrian Target Tracking Algorithm Based on Surf and Improved Correlation Filtering

Online Multipedestrian Tracking Based on Fused Detections of Millimeter Wave Radar and Vision

Intelligent Video Surveillance for Checking Attendance of Traffic Controllers in Level Crossing

Pedestrian movement trajectory reappearance and crossing feature expression based on video processing

Motion Targets Tracking Algorithm Based on Video Surveillance

Long-term visual tracking algorithm for UAVs based on kernel correlation filtering and SURF features

Representation and Analysis of Pedestrian Crossing States Based on Video Tracking

Adaptive Multi-Pedestrian Tracking by Multi-Sensor: Track-to-Track Fusion Using Monocular 3D Detection and MMW Radar

Target Detection from SAR Images Based on Wavelet Transform De-Noise and Improved CFAR

Online Multi-target Tracking for Pedestrian by Fusion of Millimeter Wave Radar and Vision

Siam Deep Feature KCF Method and Experimental Study for Pedestrian Tracking

HSKP-CF Algorithm Based on Target Tracking for Mobile Following Robot

A Real-Time Tracking Method Based on SURF

A long-memory pedestrian target tracking algorithm incorporating spatiotemporal trajectory feature enhancement model

Tracking algorithm of snowboard target in intelligent system

Effective multiple pedestrian tracking system in video surveillance with monocular stationary camera

Pedestrian tracking algorithm by video surveillance based on ensemble detection

An Improved Particle Filter Approach for Real-time Pedestrian Tracking in Surveillance Video

A Hybrid Visual Tracking Algorithm Based on SOM Network and Correlation Filter

Cross-Video Pedestrian Tracking Algorithm with a Coordinate Constraint

Tracking A Variable Number of Pedestrians in Crowded Scenes by Using Laser Range Scanners