The Model of Visual Attention Infrared Target Detection Algorithm
Tingjun Li, Zhang Fuguang, Cai Xinju, Huang Qilai, Qiang Guo · 2010
The model of visual attention infrared target detection algorithm is presented. Mainly the visual features are extracted from the brightness contrast and movement in the current frame still images and image sequences of the motion vector, and then a linear convergence significantly diagram, with locally adaptive thresholding instead of "Winner-Takes-All" neural network (Winner-Take-All, WTA), through the similarity of pixel gray scale and significant regional centroid of the adjacency to split the objectives and background, finally be interested in infrared image targets (including thermal targets and moving targets). Simulation results show that the method for the fusion system, the lower the contrast of the image after the video sequence scene moving target detection with good results.