A Robust Approach for Anti-jamming Target Tracking in Forward Looking Infrared Imagery
Guanhua Su, Huimin Ma, Yu Hou · 2011
In this paper, we present a new robust approach of target tracking in infrared imagery based on Mean Shift tracking and clustering algorithm. Our approach works robustly on jammed images, which is usually under extreme situation. We first define the Jammed forward looking by 4 Features, and proposed tracking algorithm based on them. The proposed tracking algorithm mainly references Camshift, yet has many unique features. First, a quick mean shift clustering algorithm for one-dimension data is applied to shorten the time cost. Second, we introduce the idea of double possibility distribution image to get accurate position and size of target. Third, we discard the kernel function to avoid the influence of interference. Results show that the proposed tracking approach works smoothly despite of several harmful factors which other approaches fail to overcome.