TLD target tracking algorithm based on a visual attention mechanism

Abstract

The invention relates to a TLD target tracking algorithm based on a visual attention mechanism, and belongs to the technical field of artificial intelligence. According to the invention, an original TLD target tracking algorithm is taken as a basis; the detection module of the TLD is subjected to salient region extraction of a visual attention mechanism, so that interference of a non-salient region can be eliminated; salient information can be effectively screened through visual saliency, a visual saliency map is formed, then the saliency map is classified, a detection module of a TLD algorithm is further optimized, a target model and a background model are established through an integrated classifier, the output of the integrated classifier serves as the input of a nearest neighbor classifier, and similar samples are screened out. The method is good in stability under the interference of the external environment, when the target is shielded and quickly rotated, the visual attention mechanism updates the target feature in real time, and good performance is shown.

Description

technical field [0001] The invention relates to a TLD target tracking algorithm based on a visual attention mechanism, and belongs to the technical field of artificial intelligence. technical background [0002] With the development of machine vision technology, object tracking has gradually become a promising development direction and gradually become a research hotspot. The tracking of moving objects has been maturely used in traffic assistance, robot research and development, human behavior analysis and video surveillance in the current society. Target tracking has gradually become a promising development direction, and this technology is increasingly Maturity depends on the rapid development of computer technology, and the cost of computing and storage has also been greatly reduced. my country's research on target tracking technology began in the late 1960s, and has gone through more than 50 years of growth. At present, my country's target tracking technology has made g...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the problem that target tracking tends to cause drift in complex environments and the stability of the tracking results is not high, the present invention proposes a TLD target tracking algorithm based on the visual attention mechanism based on the original TLD target tracking algorithm, which can be used when the external environment is disturbed. The lower stability is better. When the target is occluded or quickly turned, the visual attention mechanism updates the target features in real time, showing good performance. To complete the feature extraction and feature update of target tracking, the following steps are included:

Images