Year: 2016
Type: conference-paper
Abstract: Tracking-Learning-Detection(TLD) algorithm has the advantages of high speed, high accuracy of tracking rate and self-detection mechanism.However the sensitivity to illumination variation and background clutter leads to drift even miss for TLD algorithm under the complex environment.An improved TLD visual tracking algorithm based on depth information is proposed, which is improved from three aspects.The first is that a foreground extraction algorithm based on denoised depth data is adopted to extract the region of interest.Second, a sophisticated detection of blob approach is used to narrow down the searching area.And the third is that the non-maximal suppression strategies are applied to optimize the result.The experimental results show that the time complexity of proposed is decreased and the robustness is upgrade under challenging circumstances.
Authors: Jialiang Jialiang Mao, Xiaorong Xiaorong Shen
Institution: Beihang University
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Award: G1461886303
OpenAlex ID: W2324779522
